Population projection of the north Scotland Red. Kite (Milvus milvus) population

Transcription

1 Population projection of the north Scotland Red Kite (Milvus milvus) population Matthew Geary and Brian Etheridge Running head: Population projection of the north Scotland Red Kite (Milvus milvus) population Matthew Geary Department of Environmental and Geographical Sciences, Manchester Metropolitan University, Chester Street, Manchester, M1 5GD, UK. Brian Etheridge Royal Society for the Protection of Birds, Etive House, Beechwood Park, Inverness, IV2 3BW, UK. Corresponding author: Matthew Geary

2 Summary 1. The Red Kite (Milvus milvus) is a scavenging raptor found throughout Europe and native to Britain. The British population was reduced to a handful of breeding pairs during the 20 th century due largely to persecution. 2. Red Kites were reintroduced to north Scotland and southern England in 1989 with differing levels of success. The southern England population experienced more rapid population growth than the north Scotland population despite similar productivity. Illegal persecution is believed to be a major factor in this difference. 3. A stage based stochastic matrix model of the north Scotland Red Kite population was constructed from monitoring data between 1995 and This was run under a Monte Carlo simulation over 50 years for 1000 iterations. An extinction threshold was set at 30 breeding pairs and an escape threshold of 100 breeding pairs. Elasticities for the demographic parameters were calculated for the model. 4. Recruitment under different levels of persecution was estimated from monitoring data. Both a low and high estimate of persecution deaths was used. These estimates were used in stochastic matrix models to predict population growth with lower persecution levels. 5. The mean stochastic matrix model resulted in a declining population (λ=0.971, 95% C.I: ) with an extinction probability of The model using the low estimate of the effect of persecution resulted in a slowly increasing population (λ=1.015, 95% C.I: ) with a 99.8% chance of extinction and a 8.3% chance of escape. The high estimate of the effect persecution resulted in a rapidly increasing population (λ=1.107, 95% C.I.: ) with an 84.8% chance of extinction and a 30.4% chance of escape. 7. The north Scotland Red Kite population has a currently declining mean growth rate which would be most influenced by changes in the survival of breeding adults. It is therefore vital that this remains high. Models reflecting the absence of illegal persecution have an increased growth rate and achieve growth similar to the levels found in southern England.

3 Introduction Reintroductions by translocation are an increasingly used method of restoring animal populations to areas in which they have been lost due to human involvement (Griffith et al. 1989). Mammals (e.g. Arabian Oryx; Oryx leucoryx), island birds (e.g. Stitchbirds; Notiomystis cincta) and raptors (e.g. Peregrine Falcon; Falco peregrinus) have all been involved in previous reintroduction projects (Seddon et al. 2007). In Britain both the Red Kite (Milvus milvus) and White-tailed Eagle (Haliaeetus albicilla; Marren 2002) have been reintroduced to parts of their former ranges. For reintroduction projects to be successful, it is vital that a viable population of wild individuals can be achieved (Schaub et al. 2004). In order to track the fate of a reintroduced population a comprehensive monitoring programme should accompany it (Carter 2007). This occurred for the Red Kite reintroductions and it is those data which allow population analysis to be attempted today. Red Kites were once common throughout Britain (Carter 2007). During the last century their populations were reduced to just a handful of birds in the Welsh valleys by persecution (Carter 2007). In 1989, after satisfying the IUCN criteria (IUCN 1987), Red Kites were reintroduced to southern England, in the Chilterns, and northern Scotland, on the Black Isle near Inverness. A total of 93 birds were translocated from Swedish and Spanish kite populations to each of the two reintroduction sites (Evans et al. 1999). Since this time, the southern England population has increased in size to over 300 breeding pairs. The north Scotland

4 population, on the other hand, has grown slowly and reached only 46 breeding pairs in 2008 (RSPB unpublished data). This slow growth rate is believed to be due to continued illegal persecution of raptors in the north of Scotland (Carter, 2007). This has been demonstrated in several other raptor species such as the Golden Eagle (Aquila chrysaetos) (Whitfield et al. 2004) and Hen Harrier (Circus cyaneus) (Etheridge, Summers & Green 1997). The north Scotland population experiences more human-caused mortality than any other red kite population and the majority of these deaths are the result of poisoning. Lacing carcasses with poison is one illegal method of predator control on shooting estates and, as scavengers; Red Kites are particularly vulnerable to this type of activity (Carter 2007). Matrix models have long been used to study the demographics of animal, as well as plant, populations (Caswell 1989). In recent years they have been shown to be particularly useful in studying threatened populations in order to project their future growth as well as identify management objectives with minimum impact upon the population (e.g. Schaub et al. 2004, Tirpak et al. 2006). To be at their most effective, matrix models require detailed, long-term datasets and given that Red Kites on the Black Isle have been intensively monitored since 1989 this population lends itself to the matrix modelling approach. Matrix models rely on certain assumptions about a population (Case 2000) and these are met by the red kite population in question. The north Scotland population is isolated from other kite populations and, based on wingtag sightings, there has been no immigration from elsewhere. Due to the slow rate of increase, the population has not yet reached carrying capacity and no density

5 dependent effects are evident. This is indicated by the number of available territories within the core breeding range along with the continued high breeding productivity of the population. This investigation aims to assess the growth and long term viability of the north Scotland Red Kite population using a stochastic matrix model. This will be achieved using prospective analysis to ascertain which demographic parameters are most vulnerable to change as well as to recommend management strategies to ensure the survival and growth of the population. Projection under different mortality regimes based on estimates of persecution levels will be used to investigate the effect that this has on the population.

6 Materials and methods Red Kite monitoring Red Kites are systematically monitored according to the methods set out in Hardey et al. (2007) in a yearly census. Nests are located in March and April when breeding pairs can be found displaying and nest building. All nests are assumed to have been located during the years from which data was used. Between late March and early May eggs are laid and this is indicated by the presence of an incubating female. In late May nests are re-visited to check for young. A nest containing young will have faecal material, or whitewash, spread around the base of the nest tree. With experience the size and number of young present can be estimated using the quantity of faecal material (Hardey et al. 2007). In July and August the nesting sites are checked for the presence of fledged young. After the visit to check for the presence of young, a further visit follows in which the nest tree is climbed. During this visit the young are fitted with BTO metal leg rings. Along with this patagial wing tags are fitted to both wings. The colour of the left wing indicates the population from which the bird originated (blue in the case of north Scotland) and the right wing tag the year in which it was born. The tags have markings such as letters or numbers so that each kite can be identified individually. During the autumn and winter local communal roost sites are monitored at least once a week. As roosts are large and conspicuous, all roosts are assumed to have been located. Communal roosts vary in size with between ten and 80 birds present and a

7 sample of wing tags can be read using telescopes. Presence at communal roosts confirms that an individual kite is still alive. During the breeding season the wing-tags of nesting adults are similarly read and recorded. Matrix model The model is a 3x3 matrix representing three stage classes. These are juvenile birds, second years and adult birds. Birds breed for the first time in their second year reflecting the modal age of first breeding in the population. Each parameter is drawn at random from a distribution based on the mean and standard deviation of the monitoring data for that parameter. Survival parameters were based on a beta distribution and fecundity on a Gaussian distribution. The model was run under a Monte Carlo simulation for a period of 50 years with 1000 iterations to ascertain asymptotic growth rate and long term viability of the population as well as the elasticity of each demographic parameter. The Monte Carlo simulation enables a calculation of the probability of extinction and escape for the population. In this case the extinction threshold was set at 30 breeding pairs which was the starting population for the model. The escape threshold was set at 100 breeding pairs; a level which this population was initially expected to reach by All analysis was performed using the ULM software version 4.5 (Legendre & Clobert 1995). Parameter estimation All parameters were estimated from the monitoring data between 1995 and In the north Scotland population, survival of adult birds is high and not influenced by the

8 age of the bird. Fecundity of two year old birds was lower (Wilcoxon signed rank, W=4285, p=0.004)) than that of older birds as has been shown in other British populations (Carter 2007). Juvenile birds, i.e. first years, have never bred in the Scottish population so juvenile fecundity was set as 0. DNA analysis of chicks in the north of Scotland population was used to sex birds during and again in This information showed a 1:1 sex ratio for the birds which is common with other kite populations (Carter 2007). The number of fledged female young was used as the fecundity of adult birds within the model. This was taken to be half the mean number of fledged young. The mean recruitment rate of young birds into the breeding population was used in place of mortality for first year birds. This does not imply that all non breeding birds have died but merely reflects their lack of contribution to the breeding population. Due to the adverse effect which illegal persecution is thought to have on the north Scotland Red Kite population, the model was used to project the direction of population growth under different levels of persecution. Mortality of non-breeding kites was calculated using roost monitoring data. A kite was considered to have left the population when it was not found breeding and did not appear at communal roosts. This was calculated for each year between 1992 and 2002 for first and second year birds. A linear model was then used to regress the number of 0, 1 and 2 year olds considered dead with values for recruitment into the breeding population (p= 0.04, df= 10, R 2 = 0.4). Using data of confirmed deaths from , the percentage of birds killed by poisoning or shooting was calculated.

9 The effect of deaths from illegal persecution was investigated under two scenarios. The two scenarios were both designed to simulate population growth with no persecution. The first scenario was a conservative estimate; this was the known persecutions model. In this scenario the recruitment rate was increased as if those birds known to have been killed by humans were still alive. The second scenario was a high estimate of losses to persecution; this was the percentage persecutions model. In this scenario the percentage of confirmed deaths caused by humans was applied to the missing birds in the population and the recruitment rate increased as if these birds had survived. In all three projections the model was started with an arbitrary population size of 30 breeding pairs and 20 first year birds.

10 Results Parameter estimation Mean annual second-year fecundity was found to be 0.69 female young fledged per breeding female with a variance of For older birds the mean annual fecundity was 0.96 female young per breeding female with a variance of Mean annual adult survival was estimated at 85.98% with a variance of 8.72%. Mean annual juvenile recruitment was found to be 11.67% with a variance of 9.68%. Juvenile fecundity was zero. Matrix model The stochastic matrix model (Figure 1.) resulted in a declining population with λ=0.971 (95% C.I: ) which is significantly different from 1. The mean generation length of the model was 8.58 years and the net reproductive rate per annum. The probability of extinction (i.e. decreasing to below 30 breeding pairs) was 1. Elasticity Adult survival had the highest elasticity at Juvenile recruitment was next with followed by adult fecundity at Second-year fecundity had the lowest elasticity with <0.001.

11 Projections under different recruitment scenarios The 'known persecutions' model had a recruitment rate of 17.2%. The 'percentage persecutions' model had a recruitment rate of 30.9%. When used in the model as the mean of a beta distribution with the same variance as the actual recruitment value, and keeping all other parameters constant, the following values of λ were found: The known persecutions model (Figure 2.) had λ=1.015 (95% C.I: ) which is significantly greater than one. This represents a slowly increasing population. The percentage persecutions model (Figure 3.) had λ=1.107 (95% C.I.: ) which is significantly greater than one. This indicates an increasing population. Probabilities of extinction were for the known persecutions model and for the percentage persecutions model. Probabilities of escape were for the known persecutions model and for the percentage persecutions model.

12 Discussion The model of the north Scotland Red Kite population gave a negative stochastic growth rate (λ) which is significantly different from 1. This suggests that the population will decrease over the next 50 years. The growth rate is lower than the previous estimate for the Black Isle population (Evans et al. 1999) which looked at a smaller data set. It is also lower than estimates for other Red Kite populations (Evans et al & Carter et al. 1999). These results indicate that the population is unlikely to be viable in the long term if conditions remain the same. Survival is the most difficult parameter to estimate as it is difficult to be sure a bird has died without finding a corpse or ring. However, in a marked population with predictable behaviour patterns such as fidelity to nesting territory and winter communal roosts, this problem is reduced. Even so, some kites are not wing-tagged each year due to fledging before they are tagged or nest being inaccessible (2 nests in 2008) and their fates are therefore unknown. However this is likely to be a small percentage (less than six birds per year) of the population and therefore should not affect the estimates of survival. Missing birds (i.e. those which disappear from the population) do not represent a problem in terms of estimating survival they are not recruited into the breeding population, nor are they replaced by immigration, so it is not necessary to include them in the model. The demographic parameters used in the model compare favourably with other Red Kite populations as well as those of other raptors. Evans et al. (1999) found fledged

13 brood sizes of 1.9 for southern England and 1.6 for northern Scotland which is similar to the adult fecundity used in the model which was 1.9 chicks per female. When second-year fecundity is also considered the average number of chicks per female is 1.7 which is higher than this original estimate for north Scotland.. It is somewhat lower than the value of 2.1 used for modelling the removal of kites from the southern England population for another reintroduction by Carter et al. (1999), but greater than that reported for the Welsh population by Davis, Cross and Davis (2001) which was 1.19 to 1.4 chicks per successful nest. The mean adult survival was 85.98% which is lower than that used by Evans et al. (1999) as well as the 95% reported by Carter (2007) for both the southern English and Welsh populations but fits into the range of survival values reported by Real and Mañosa (1996) for long lived raptors. Elasticities are the most commonly used perturbation analysis for conservation. Other studies have used the rankings of these to inform management decisions for threatened species and they can be used to identify important demographic parameters (Benton & Grant 1999). In this case the elasticities are as expected for a slow breeding raptor population with adult survival being the most influential demographic parameter (Real and Mañosa 1996). The results from this model suggest that maintaining high adult survival is most important to ensure a healthy Red Kite population with a relative contribution to λ of 86%. Juvenile recruitment has an elasticity of When considering elasticities in terms of conservation it is important to look at the wider picture and not necessarily to focus on only the largest value (De Kroon, Groenendael & Ehrlen 2000). In this case the adult survival is already high and so should be maintained. This high elasticity warns us of the problems the population would face with lower adult survival. However, improving

14 the juvenile recruitment of the population would benefit the population by having a positive effect on the growth rate and, in theory, would be easiest to achieve. The estimates of recruitment under different mortality scenarios are based on the relationship between mortality of first and second year birds and recruitment. This relationship is statistically significant (p=0.04) but somewhat weak (R 2 =0.4). The small decrease in mortality from the known persecutions model increases recruitment by a small amount which is not an unreasonable assumption. This leads to a higher growth rate which represents and increasing population. The recruitment estimate from the percentage persecutions model is much higher. This estimate represents population growth that would be expected if there was a complete absence of persecution related deaths. This estimate is based on the assumption that, as it is an illegal activity, most persecutions are not reported and therefore go unrecorded (Carter 2007). This was considered to be a high estimate of the level of persecution and so a high estimate of potential recruitment, however, it is actually slightly lower than the recruitment rate in the welsh population reported by Davis et al. (2001) and represents a smaller level of change from the mean recruitment value than that expected by Whitfield et al. (2008) for Golden Eagle populations (10% to 40%) if persecution were to be stopped. In light of these other estimates of recruitment rate this estimate seems to be more realistic for recruitment in the absence of persecution. The growth rate predicted by this model is significantly greater than 1 and represents an increasing population. In fact, represented graphically, this growth rate resembles exponential growth. This is closer to what

15 would be expected from a population with no natural predators and no density dependent effects (Lack 1967). The probability of extinction for the model of the current population is 1. This suggests that this population is unlikely to ever reach 100 breeding pairs under the current conditions. The models under reduced persecution had lower probabilities of extinction although for the known persecutions model this was still high at This result suggests that even under a reduced level of mortality the population still has a strong chance of decreasing due to environmental and demographic stochasticity. The probability of escape for the known persecutions model was which suggests that even with an improved growth rate the population is unlikely to reach 100 pairs in 50 years. Under the percentage persecutions model the population fared better with only a 84.8% chance of resulting in less than 30 breeding pairs and a 30.4% chance of reaching more than 100 breeding pairs after 50 years. These results suggest that the constraining factor for Red Kite population growth on the Black Isle is low juvenile recruitment caused by the illegal persecution of raptors. This effect has also been shown in other raptor populations and was considered one of the major causes in preventing the Golden Eagle reaching favourable conservation status in a recent Scottish Natural Heritage report (Whitfield et al. 2008). Juvenile kites are more mobile than adult birds, which tend to remain on their breeding territories, and so are more likely to be killed by persecution which is uncommon in the core breeding area.

16 The Red Kite population on the Black Isle offers many opportunities for future research. Little is known about the movements of young birds between their fledging and first breeding. Young birds generally move in a south-westerly direction (Evans et al. 1999) but the exact locations of these movements is unknown. As this dispersal period is believed to be when young kites are most vulnerable, monitoring their movements would be an extremely valuable exercise. Using satellite tags these, as well as adult birds could be monitored and their fates recorded (e.g. Yamaguchi et al. 2008). This investigation highlights several key ideas for avian reintroductions. The first of these is the need for a dedicated monitoring programme coupled with a focussed research agenda (Armstrong & Seddon 2007). Without the Red Kite monitoring programme it would be impossible to investigate the causes of slow population growth and it is debatable whether this slow growth would be detected at all. As the population is still far from safe this monitoring programme should continue. The research agenda should focus on the questions most relevant to the population reintroduction but also attempt to answer questions which might be of benefit to similar reintroduction attempts (Armstrong & Seddon 2007). In this case, the success of Red Kite reintroductions in northern Scotland and southern England led to further Red Kite reintroduction projects throughout the country (Carter 2007). Problems

17 associated with the north Scotland population as well as more detailed modelling in the future will help to inform the strategy and management of future reintroductions. In conclusion the north Scotland Red Kite population has a declining growth rate found by stochastic matrix modelling. This population is appears to be strongly affected by a low recruitment. Modelling with a reduced mortality of young birds corresponding to an estimate of human-caused deaths results in a rapidly increasing population. This supports the idea that the low recruitment rate of this population is caused by the illegal persecution of raptors (Carter 2007). Removal of this threat to the population by stronger legislation and enforcement should ensure the long term survival and growth of the population. If this threat is to continue then the focus for management should be to maintain high adult survival and high fledging rates in the population. The reintroduction of Red Kites has been a success story in Britain with a 447% increase in numbers between 1994 and 2004 (RSPB 2007). However, this overall growth hides the threats which the north Scotland population faces and will have to overcome in order to achieve long term persistence. If raptors, including the Red Kite, continue to be killed illegally then there is no guarantee of the survival of this, or any other, population.

18 References Armstrong, D.P. & Seddon, P.J. (2007) Directions in reintroduction biology. Trends in Ecology and Evolution, 23 (1): Benton, T.G. & Grant, A. (1999) Elasticity analysis as an important tool in evolutionary and population ecology. Trends in Ecology and Evolution, 14(12): Carter, I., McQuaid, M., Snell, N. & Stevens, P. (1999) The red kite (Milvus milvus) reintroduction project: Modelling the impact of translocating kite young within England. Journal of Raptor Research, 33(3): Carter, I. (2007). The Red Kite 2nd ed., Shrewsbury: Arlequin Press. Case, T.J. (2000) An illustrated guide to theoretical ecology, Oxford: Oxford University Press. Caswell, H. (1989) Matrix Population Models, Sunderland Massachusetts: Sinauer Associates Inc. Davis, P., Cross, T. & Davis, J. (2001) Movement, settlement, breeding and survival of red kites (Milvus milvus) marked in Wales. Welsh Birds, 3: De Kroon, H., van Groenendael, J. & Ehrlen, J. (2000) Elasticities: a review of methods and model limitations. Ecology, 81(3):

19 Etheridge, B., Summers, R.W. & Green, R.E. (1997). The effects of illegal killing and destruction of nests by humans on the population dynamics of the hen harrier Circus cyaneus in Scotland. Journal of Applied Ecology, 34: Evans, I.M., Summers, R., O Toole, L., Orr-Ewing, D., Evans, R., Snell, N. & Smith, J. (1999) Evaluating the success of translocating red kites Milvus milvus to the U.K. Bird Study, 46: Griffith, B., Scott, J.M., Carpenter, J.W. & Reed, C. (1989) Translocation as a species conservation tool: status and strategy. Science, 245 (4917): Hardey, J., Crick, H., Wernham, C., Riley, H., Etheridge, B. &Thompson, D. (2007). Raptors: A Field Guide to Survey and Monitoring, Edinburgh: The Stationery Office. IUCN (1995) Guideline for re-introductions (Accessed 2008) Lack, D. (1967) The natural regulation of animal numbers, Oxford: The Clarendon Press. Legendre, S. & Clobert, J. (1995). ULM, a software for conservation and evolutionary biologists. Journal of Applied Statistics, 22: (Version 4.4 downloaded from Marren, P. (2002) Nature Conservation, London: HarperCollins. Real, J. & Mañosa, S. (1996) Demography and conservation of western European Bonelli s eagle Hieraaetus fasciatus populations. Biological Conservation, 79: 59-66

20 Redfern, C.P.F. & Clark, J.A. (2001) Ringer s Manual, Thetford: British Trust for Ornithology. RSPB (2007) Birds of prey in the U.K.: on a wing and a prayer. Schaub, M., Pradel, R. & Lebreton, J-D. (2004) Is the reintroduced white stork (Ciconia ciconia) population in Switzerland self-sustainable? Biological Conservation, 119: Seddon, P.J., Armstrong, D.P. & Maloney, R.F. (2007) Developing the science of reintroduction biology. Conservation Biology, 21(2): Tirpak, J.M., Giuliano, W.M., Miller, C.A., Allen, T.J., Bittner, S., Buehler, B.A., Edwards, J.W., Harper, C.A., Igo, W.K., Norman, G.W., Seamster, M. & Stauffer, D.F. (2006) Ruffed grouse population dynamics in the central and southern Appalachians. Biological Conservation, 133: Whitfield, D.P., Fielding, A.H., McLeod, D.R.A. & Haworth, P.F. (2004). Modelling the effects of prersecution of the population dynamics of golden eagles in Scotland. Biological Conservation, 119: Whitfield, D.P., Fielding, A.H., McLeod, D.R.A. & Haworth, P.F. (2008) A conservation framework for golden eagles: implications for their conservation and management in Scotland. Scottish Natural Heritage commissioned report No. 193 (ROAME No. F05AC306).

21 Yamaguchi, N., Tokita, K-I., Uematsu, A., Kuno, K., Saeki, M., Hiraoka, E., Uchida, K., Hotta, M., Nakayama, F., Takahashi, M., Nakamura, H. & Higuchi, H. (2008) The large-scale detoured migration route and the shifting pattern of migration in Oriental honey-buzzards breeding in Japan. Journal of Zoology, 276: 54-62

22 Figure Legends Figure 1. Mean number of Red Kite breeding pairs in the north Scotland population from a Monte Carlo simulation of the stochastic matrix model over 50 years with 100 repetitions. The population is declining with λ=0.971 (95% C.I: ). 95% confidence intervals are shown as dotted lines. Figure 2. Mean number of Red Kite breeding pairs in the north Scotland population from a Monte Carlo simulation of the known persecutions stochastic matrix model over 50 years with 100 repetitions. The model gives a slowly increasing population with λ=1.015 (95% C.I: ). 95% confidence intervals are shown as dotted lines. Figure 3. Mean number of Red Kite breeding pairs in the north Scotland population from a Monte Carlo simulation of the known persecutions stochastic matrix model over 50 years with 100 repetitions. This is a rapidly increasing population with λ=1.107 (95% C.I.: ). 95% confidence intervals are shown as dotted lines.

23 Figures Figure 1

24 Figure 2.

25 Figure 3.