A baseline survey of Lesser Kestrel Falco naumanni in south-east Kazakhstan

STEPHEN J. PARR, SERGEI SKLYARENKO, SERGEI BROKHOVICH, JANE BROOKHOUSE, PAUL N. COLLIN AND BORJA HEREDIA

The Lesser Kestrel Falco naumanni population in South-east Kazakhstan is estimated, from a random sample survey of 40 10-km squares, at 500-2000 pairs. Breeding colonies were found most commonly in rock cliffs, but also in sand cliffs along rivers. Colony presence was associated with presence of rock cliffs, grass steppe and maximum altitude of 1000-2000 metres. Maximum attitude was easier to model than grass steppe or the presence of rock cliffs because of the limitations of available maps. A maximum altitude model predicted a population of c. 2000 pairs, close to the upper 95% confidence limit of the extra polated estimate. The population was apparently unthreatened by land-use changes, as agriculture in the region was depressed and abandoned in marginal habitats. The contrast with central Turkey and southern Spain where colonies are associated with flat topography, threatened by intensive non-rotational agriculture and nests located in cavities within buildings is discussed.

 

INTRODUCTION
	LESSER KESTREL Falco naumanni is a small falcon that breeds colonially in buildings, or in cliffs, and hunts over open country. It was formerly widely distributed across the Palearctic south of 55oN. It is currently classified as globally threatened because of widespread population declines related to habitat change and pesticide use in the Western Palearctic (Collar et al. 1994). Its status in the Eastern Palearctic is poorly known, but it was recently estimated that the Republic of Kazakhstan holds more than 10,000 pairs, making it the largest national population in the world (Zollinger & Hagemeijer 1994). The International Action Plan (Biber 1996) recommended that surveys be undertaken to establish its status and identify key breeding concentrations in this country. Steppe and dry grassland are important foraging habitats (Parr et al. 1997), of which Kazakhstan holds a very large and important resource (Walter & Box 1983). The economic and agricultural reforms initiated with the formation of the Republic of Kazakhstan in 1991, following the collapse of the Soviet Union, make this an opportune time to conduct a baseline survey. Future, repeat surveys will provide a means to monitor population trends, and provide an indication of the health of open-country habitats. Zollinger & Hagemeijer (1994) identified five population concentrations in Kazakhstan. The largest stretched across the north of the country and approximately corresponded with the dry steppe vegetation zone of Walter & Box (1983). The other four smaller concentrations corresponded k) areas of montane steppe in the east and south-east. The study area Included two of these, south and south-east of Lake Balkhash (Fig. 1) and covered an area of c. 158,000 km2, approximately 6% of the country.
Figure 1	Figure 1. Diagram of 158,000 km2 study area in south-east Kazakhstan. The area surveyed excluded land over 3,000 metres and water bodies.
METHODS
	The project was a partnership between two scientists from the Institute of Zoology in Almaty and four European ecologists with experience of similar work in Turkey and Spain. The survey team used two 4-WD vehicles together with two experienced drivers. The methods mirrored those employed to produce a population estimate for central Turkey (Parr et al. 1995). A 5% random sample of all open country 10-km squares-high mountains and water bodies being excluded-was attempted using the Universal Transverse Mercator grid from US Tactical Pilotage Charts. This produced a total of 79 40-km squares in the survey area. In south-east Kazakhstan, Lesser Kestrel is known to breed in natural sites, either in rock cliffs or in sand cliffs along rivers (S. Sklyarenko pers. obs.), whereas in Turkey they breed under roof tiles and other cavities within settlements (Parr et al. 1995). Hence, all potential rock and river cliff sites within randomly selected 10-km squares were checked for the presence of colonies. Nine habitat types: grassland, Artemisia spp. steppe, sand/stone desert, sand dune, wetland, upland scrub, arable agriculture, long-term fallow and poor fallow, were recorded as either present or absent within 10-km squares, as measuring areas proved difficult and time-consuming because of the low density of tracks. Travel overland was possible but slow. Three other variables, mean altitude, maximum altitude and topography score were measured from the US Tactical Pilotage Charts. A car-mounted GPS system (Magellan 2000 plus adapter) was used to locate each 10-km square and thence plot the survey route on 1:500,000 maps US Tactical Pilotage Charts as no accurate, smaller-scale maps were available and there was a general lack of spatial reference points in the landscape. Forty randomly selected 10-km squares were checked between 26 April and 20 May 1997. The mean length of time spent in each square was 2.8 ± 0.4 hours. Thirty-nine squares were not surveyed because of lack of time. At the start of the fieldwork in the south-east of the survey area, all squares within UTM grid squares were checked to complete the 5% sample (squares DJ, EJ and FJ). It swiftly became apparent, from checking squares further north in desert and wetland habitats (squares DK, EK and FK), that these flat, open-country areas contained no available cliff nest sites. Thus coverage in these areas, which were clearly defined on the US Tactical Pilotage Charts, was reduced and survey effort concentrated on 40-km squares lying adjacent to mountain ranges, as well as squares that contained low hills with potential nest sites. Hence, 34 of the 39 unsurveyed 40-km squares were within desert, sand-dune or wetland habitats and most probably did not hold breeding Lesser Kestrels. Overall coverage of potential breeding habitats in the region was close to 5%.
RESULTS
	Lesser Kestrel was recorded in ten (25%) of the 40 squares checked. In two 10-km squares, flocks were recorded hunting over open country and the presumed nest sites were in adjacent 10-km squares. In the remainder, birds were recorded at, or adjacent to, cliff nesting sites and were classified as breeding. Colonies in rock cliffs on the edges of mountains or hills were located in a total of 14 10-km squares, seven of which were located in non-random squares and one 10-km square held colonies of Lesser Kestrels within sand and rock cliffs along a river. The mean flock size within random squares was 12.1 ± 2.2, with a maximum of 29. Extrapolation from the 5% sample of open-country habitats, assuming that all unsurveyed squares did not hold colonies, gave an estimate for the survey area of 2600 individuals (Table 1) within only 62,000 km2 (40%) of the 158,000 km2 survey area. Three 100-km squares - DJ, KP and MS - held more than 500 birds each. 95% confidence limits were calculated, which gave a population estimate of 2600 + 1470 individuals. Assuming all birds observed were breeding gave an estimate of c.500-2000 breeding pairs. Table 1. Summary of Lesser Kestrel Falco naumanni (LK) data collected in south-east Kazakhstan, April-May 1997. 100-km square 5% sampIe* Number surveyed Number occupied LK number LK population estimate DJ 5 5 2 32 640 DK 5 3 0 0 0 EJ 5 5 2 13 260 EK 5 2 0 0 0 EL 5 0     0 EM 2 0     0 FJ 5 5 1 14 280 FK 5 2 0 0 0 FL 5 0     0 FM 2 0     0 GJ 2 1 0 0 0 GK 2 2 0 0 0 GL 1 0     0 GM 1 0     0 KP 2 2 2 28 560 KQ 1 1 0   0 KR** 2 0     0+ LP** 5 1 1 9 180+ LQ 1 1 0 0 0 LR 5 2 0 0 0 LS 4 2 0 0 0 MR 4 2 1 5 100 MS 5 4 1 29 580 TOTAL 79 40 10 130 2,600+ * 5% sample of open country habitats **one square in KR and four in LP held potential nesting habitat but the remaining unchecked squares were all within the desert or sand-dune habitats and were unlikely to support colonies Individual logistic regression models were constructed for each independent variable, concerning the presence or absence of Lesser Kestrel, to best establish any associations with habitats or other variables, because the sample of occupied 10-km squares by breeding colonies (n = 8) was so low. This revealed that four measures - topography, mean altitude, maximum altitude and the presence of grass steppe - exhibited a significant positive correlation with the presence of breeding Lesser Kestrel (Table 2). A matrix of product-moment coefficients demonstrated that all these measures were significantly inter-correlated and also that grass steppe was significantly negatively correlated with desert and wetland habitats (Table 3). These results confirmed that breeding Lesser Kestrel in south-east Kazakhstan are restricted to nesting in areas with cliff nest sites. These are predominantly in low hills or on the edge of mountain ranges. Table 2. Logistic regression results of tour significant independent variables on Lesser Kestrel Falco naumanni presence. Variable Estimate SE T-ratio P Max. Altitude 0.001 0 2.476 0.013 Altitude 0.001 0 2.243 0.025 Topography 0.229 0.081 2.834 0.005 Grass Steppe 3.045 1.144 2.661 0.008 Table 3. Matrix of product-moment coefficients of 14 independent variables. (correlations significant at p <0.05 in bold) Altitude Arable Art. St Desert Fallow Grass St Max. Alt P. Fallow Rock Cl S. Dune Sand Cl Topog U. Scrub Wetland 1.000                           -0.116 1.000   0.319 -0.357 1.000   -0.300 -0.352 0.020 1.000   -0.199 0.799 -0.156 -0.352 1.000   0.588 0.013 0.182 -0.357 0.013 1.000   0.949 -0.215 0.289 -0.160 -0.295 0.533 1.000   -0.006 0.150 -0.034 -0.154 0.317 0.086 -0.089 1.000   0.638 -0.037 0.137 -0.200 -0.037 0.623 0.636 -0.070 1.000   -0.378 -0.221 -0.245 -0.039 -0.352 -0357 -0.382 -0.154 -0.338 1.000   0.214 -0.207 0.006 -0.091 -0.207 0.077 0.294 0.020 0.232 -0.091 1.000   0.842 -0.251 0.225 -0.260 -0.284 0.503 0.845 -0.122 0.628 -0.260 0.510 1.000   0.650 -0~246 0.190 -0.174 -0.246 0.488 0.653 -0.126 0.515 -0.174 0.158 0656 1.000   -0.378 0.306 -0.377 0.134 0.175 -0.085 -0.312 0.066 -0.062 -0.039 0.067 -0.260 -0.174 1.000 Altitude Mean altitude measured from 4 corners of 10-km square Arable Arable agriculture Art St Artemisia-dominated steppe Desert Sand/stone desert Fallow Long term fallow land Grass St Grass-dominated steppe Max. Alt Maximum altitude within 10-km square P. Fallow Poor fallow land Rock cliff Presence of rock cliff S.Dune Sand dune dominated by scrub Sand cliff Presence of sand cliff along river channel Topog Topography score: number of contour intersections along 10-km square boundary U. Scrub Upland scrub habitats Wetland All wetland and riverine habitats Normal regression analysis of the same four variables on the square root of Lesser Kestrel breeding numbers demonstrated that all were similarly significantly related. The same analysis, including the non-random sample, proved that the presence of grass steppe was the best predictor (r2 = 27.0, d.f. = 46, p = 0.001). There were no available vegetation maps for the study area and hence no means of mapping the extent of grass steppe and predicting the population size using the regression model. However, maximum altitude was an easily measured attribute and a line fitted by distance-weighted least squares (a weighted quadratic regression) produced a non-linear model that most usefully explains the observed distribution of breeding colonies (r2 = 20.6, d.f. = 46, p<0.01). The model suggested that colonies are be absent from low-lying desert, sand-dune and wetland habitats, as these held no cliff breeding sites and were also absent from mountains with a maximum altitude over 2000 metres. The model thus suggested that colonies were most likely in low hills and small mountain ranges, or on the edge of the Tien Shan Mountains, between 1000 and 2000 metres. The model however, could not predict the presence of colonies within cliffs along rivers. Table 4. Predicted Lesser Kestrel Falco naumanni (LK) population size from an altitude model. 100-km square MAXALT* between 1000-2000 metres Predicted LK population** DJ 55 660 DK 0 0 EJ 40 480 EK 0 0 EL 0 0 EM 0 0 FJ 5 60 FK 10 120 FL 0 0 FM 0 0 GJ 15 180 GK 20 240 GL 0 0 GM 0 0 KP 15 180 KQ 25 300 KR 5 60 LP 40 480 LQ 30 360 LR 25 300 LS 0 0 MR 30 360 MS 5 60 TOTAL 320 3,840 * MAXALT = number of 10-km squares with maximum altitude of 1000-2000 metres ** Extrapolation based on the mean recorded figure of 12 Lesser Kestrels/10-km square This model was used to predict Lesser Kestrel population size by using the mean colony size of 12 individuals/10-km square. This produced a total of 3840 individuals, or nearly 2000 pairs (Table 4). This was within the upper 95% confidence limits of the extrapolated population, which is reasonable given the lack of survey coverage in squares KR and LP, which held potential breeding habitats. However, close agreement did not exist between estimates for individual UTM squares, except for square DJ (Fig. 2). Square LP received incomplete coverage and contained potential breeding habitat as the altitude model correctly predicts. UTM squares FJ, KP and MS held large extrapolated populations but only contained small areas of suitable habitat, reflecting the small sample size from which the extrapolations were made.
Figure 2	Figure 2. Relationship between altitude model and extra-polated estimates for Lesser Kestrel Falco naumanni population size. LP** was incompletely surveyed.
DISCUSSION
	Lesser Kestrel in south-east Kazakhstan is widely distributed on the edges of mountain ranges but absent from desert, sand-dune and wetland habitats. This is best explained by the need for cliff nest sites in either mountains or river valleys. The extrapolated population estimate was 1300 ± 750 pairs, compared to a model of presence related to the maximum altitude of a 10-km square, which predicted a population of c. 2000 pairs. The difference is probably due to the small sample of 40-km squares surveyed. The best means of identifying the presence of colonies would be to map all the available rock cliff and river cliff habitats and sample these for breeding Lesser Kestrel. In the absence of such maps, further sampling of land between 1000 and 2000 metres would be valuable to confirm the model and produce a better population estimate. If the number of pairs in two of the population concentrations identified by Zollinger & Hagemeijer (1994) held up to 2000 pairs, then the Kazakhstan population is perhaps only 5000-8000 pairs. Surveys of the other three areas identified by Zollinger & Hagemeijer (1994) would be valuable to confirm other population concentrations, increase baseline data and improve the model.
Figure 2	Figure 2. Lesser Kestrel by Mike Langman.
	How far, geographically, the altitude model might be extrapolated, both within Kazakhstan and also west to the Ukraine and the Caucasus and east to north-east China where an isolated population occurs (Meyer de Schauensee 1984) is unknown. Del Hoyo et al. (1994) do not suggest that breeding in buildings and natural cliff sites occurs in geographically isolated parts of the breeding range. However, data from south-east Kazakhstan, central Turkey (Parr et al. 1995) and Spain (Bustamente 1997, Forero et al. 1996) suggest that Lesser Kestrel either uses buildings or cliffs. This might reflect the distribution of suitable feeding habitats. In Spain and Turkey, Lesser Kestrel breeds in buildings and forages over dry grasslands and agricultural land in areas of flat topography (Donázar et al. 1993, Parr et al. 1995, Bustamente 1997, Parr et al. 1997). In south-east Kazakhstan, grass steppe habitats, over which hunting birds were most often observed, is restricted to areas close to mountain ranges above 1000 metres. However, hypotheses concerning nest-site selection and foraging preferences must also take account of the differing architecture and settlement density between the countries. in south-east Kazakhstan, no villages were observed with clay-tiled roofs as in central Turkey and much of Spain. Nearly all roofs were constructed from corrugated iron and did not appear to offer suitable nesting sites. Also settlement density in south-east Kazakhstan is far lower, and less than 100 years ago were almost entirely absent since the people were nomadic, than in central Turkey and hence cliff nest sites were the only sites available in many areas of south-east Kazakhstan. The presence of such large areas of long-term and poor fallow land within the survey area provides a measure of the abandonment of agriculture, especially in marginal, upland habitats, following the collapse of the Soviet Union in 1991. The populations of other readily observed species, dependent on dry grasslands, such as larks, shrikes Lanius spp., European Bee-eater Merops apiaster and European Roller Coracias garrulus, were also abundant and apparently non-threatened.
ACKNOWLEDGEMENTS
	We thank our drivers Sergei and Ura for their skills and endurance. Funding was kindly provided by British Petroleum) Kazakhstan Ltd., the Royal Geographic Society (the David Cross Expedition Fund), the Royal Society for the Protection of Birds and the British Ornithologists' Union. Advice was gratefully received from Richard Porter and John O'Sullivan. David Moore of British Petroleum made the survey possible.
REFERENCES
	BIBER, J.-P. (1996) Action Plan for the Lesser Kestrel Falco naumanni. In: Heredia, B., Rose, L. and Painter, M. (eds.) Action Plans for Globally Threatened Birds in Europe. Council of Europe, Strasbourg. BUSTAMENTE, J. (1997) Predictive models for Lesser Kestrel Falco naumanni distribution, abundance and extinction in southern Spain. Biol. Cons. 80: 153-160. COLLAR, N. J., CROSBY, M. J. AND STATTERSFIELD, A. J. (1994) Birds to watch 2: the world list of threatened birds. BirdLife International (BirdLife Conservation Series no. 4), Cambridge. DEL HOYO, J., ELLIOT, A. AND SARGATAL, J. (EDS) (1994) Handbook of the birds of the world. Vol. 2. Lynx Edicions, Barcelona. DONÁZAR, J. A., NEGRO, J. J. AND HIRALDO, F. (1993) Foraging habitat selection, land-use changes and population decline in the Lesser Kestrel Falco naumanni. J. Appl. Ecol. 30: 515-522. FORERO, M. G., TELLA, J. L., DONÁZAR, J. A. AND HIRALDO, F. (1996) Can interspecific competition and nest site availability explain the decrease in Lesser Kestrel Falco naumanni populations? Biol. Conserv. 78: 289-293. MEYER DE SCHAUENSEE, R (1984) The birds of China. Smithsonian Institution Press, Washington DC. PARR, S. J., COLLIN, P., SILK, S., WILBRAHAM, J., WILLIAMS, N.P. AND YARAR, M. (1995) A baseline survey of Lesser Kestrels Falco naumanni in central Turkey. Biol. Conserv. 72: 45-53. PARR, S., J., NAVESO, M. A. AND YARAR, M. (1997) Habitat and potential prey surrounding Lesser Kestrel Falco naumanni colonies in central Turkey. Biol. Conserv. 79: 309-312. WALTER, H. AND BOX, E. O. (1983) Semi-deserts and deserts of central Kazakhstan. In: West, N. D. (ed.) Ecosystems of the world. Vol. 5. Elsevier, Amsterdam. ZOLLINGER, R. AND HAGEMEIJER, W. J. M. (1994) The Lesser Kestrel Falco naumanni review of the status of a globally threatened species. In: Meyburg, B.-U. and Chancellor, R. D. (eds.) Raptor conservation today. WWGBP/Pica Press, Robertsbridge. Stephen J. Parr, Jane Brookhouse and Paul Collin, Royal Society for the Protection of Birds, The Lodge, Sandy, Bedfordshire, SG19 2DL. U. K. Sergei Sklyarenko and Sergei Brokhovich Institute of Zoology, Academgorodok, 480032 Almaty, Republic of Kazakhstan. Borja Heredia, Direccion General para la Conservacion de la Naturaleza, Gran Via de San Francisco 5, 28005 Madrid, Spain.