Diet selection of lesser white-fronted geese Anser erythropus at a spring staging area

ECOGRAPHY 26: 705 714, 2003 Diet selection of lesser white-fronted geese Anser erythropus at a spring staging area Juha Markkola, Marika Niemelä and Seppo Rytkönen Markkola, J., Niemelä, M. and Rytkönen S. 2003. Diet selection of lesser whitefronted geese Anser erythropus at a spring staging area. Ecography 26: 705 714. We studied diet selection of the globally threatened lesser white-fronted goose Anser eythropus at a spring staging area on the island of Hailuoto (64 00 N, 24 45 E), off the western coast of Finland. We determined the diet using droppings, which were collected in late May, when the geese had left the area and migrated northwards. The sample potentially comprised of ejecta from 31 different individuals. Plant identification was based on visual determination of epidermal fragments. A total of 100 droppings were sampled using a point quadrat method. We calculated the percentage of each identified taxon in each dropping and related this to the availability of the corresponding taxon in the meadow. We measured preference for each taxon using Chesson s electivity index ( i ) and tested them by bootstrap resampling. The diet contained 9 taxa of the ca 40 available. Almost all dietary items were Monocotyledons (99.9%), mostly grasses (88.7%) including Festuca rubra (43%), Phragmites australis (30%) and Calamagrostis stricta (13%). Only Phragmites ( =0.73, p= 0.000), Festuca ( =0.52, p=0.004) and possibly Triglochin palustris ( =0.70, p= 0.125) were preferred, all other species were avoided. All preferred species were quite common and other goose species exploit them too. The lesser white-fronted geese preferred large natural meadows that were five times broader than an average Bothnian Bay meadow. All forms of mowing and grazing management benefit the restoration of lesser white-fronted goose habitats at the landscape level. Festuca and especially Triglochin benefit from such management. Reeds, Phragmites, whose spread has been the main cause of coastal meadow deterioration, can be controlled by management, but can also be maintained among other vegetation if mowing is less frequent or grazing not too intensive. J. Markkola ( juha.markkola@oulu.fi), M. Niemelä, and S. Rytkönen, Dept of Biology, Uni. of Oulu, P.O. Box 3000, FIN-90014 Oulun yliopisto, Finland. The lesser white-fronted goose (abbreviated later as LWfG) Anser erythropus L. was a numerous breeding bird species in arctic and sub-arctic areas between Scandinavia and the Far East before World War II (Lorentsen et al. 1999). In recent decades, population numbers have fallen and the distribution range has contracted drastically (Norderhaug and Norderhaug 1984). The most serious threat, based on results of ringing and satellite telemetry studies, is illegal or, in some countries, legal hunting (Tolvanen and Markkola 1998). At present, the autumn population comprises little over 25 000 individuals, and the species is globally threatened. In Nordic countries the LWfG population is on the verge of extinction, comprising 50 breeding pairs (e.g. Aarvak et al. 2001). Beginning from Nordic initiatives in the 1970s and 1980s a network of conservation activities was established step by step throughout the entire distribution area of the LWfG from Europe to China and the Far East. The LWfG Task Force was established under the Goose Specialist Group of Wetlands International in 1994. An action plan to protect the LWfG was approved by the European Council and published by BirdLife International (Madsen 1996). Results from a range of studies are required to construct a basis for undertaking a comprehensive vul- Accepted 24 March 2003 Copyright ECOGRAPHY 2003 ISSN 0906-7590 ECOGRAPHY 26:6 (2003) 705

nerability analysis (Soulé 1986), necessary for successful formulation of conservation policy. The conservation biology studies of recent years have included analysis of population trends and distribution (e.g. Iwabuchi et al. 1997), migration routes (e.g. Lorentsen et al. 1998) and population genetics (Ruokonen 2001) of the LWfG. One fundamental need is an understanding of the diet and habitat selection in staging areas. In particular, can an understanding of these be of benefit to the LWfG in these areas? Depending on the migration patterns and phenology of a particular bird species, conditions on the wintering quarters, in spring-staging places and in breeding areas contribute to varying degrees to the ultimate reproductive success of a population (e.g. Nilsson 1979). In northern breeding, long-distance migrating geese as the LWfG, the feeding conditions at stop-over sites are important (Ebbinge et al. 1982, Prop and Deerenberg 1991). The female geese acquire energy stores in spring-staging areas before moving to the breeding places, where very little food is available during the egg laying period. This study aimed to reveal diet and habitat preferences of the LWfG on the Bothnian Bay coast during spring staging. The results were applied to recommendations for management planning of the coastal meadows along the Bothnian Bay, with particular emphasis on recommendations for management of mowing or grazing. Material and methods Study area Diet selection of the lesser white-fronted goose was studied on coastal meadows at Tömppä on the island of Hailuoto (64 00 N, 24 45 E), on the Finnish western coast of the Bothnian Bay. During 1985 2002 this meadow supported, in most years, the majority of all lesser white-fronted geese using the Bothnian Bay flyway from the border area between Greece and Turkey via Hungary and Estonia towards Lapland and Finnmark, in northern Norway (Aarvak et al. 1999). In 1985 1988 the Bothnian Bay meadows hosted ca 100 LWfG. After that the numbers declined and in 1993, when the dropping samples were collected, 31 LWfG used the meadow of Tömppä, out of all 33 staging on the Bothnian Bay coast. In 1994 the total number was 46 but declined gradually to ca 25 in 2000 (Markkola 2001). The Valdak marshes in Finnmark, N Norway unite more branches of migration routes and therefore host more LWfG, e.g. 84 individuals in spring 1998. The central, relatively uniform part of the meadow is ca 1 km 2 and is flanked by extensive mud-flats and shallows, the area of which varies greatly according to the sea water level. Vegetation near the shoreline is dominated by Eleocharis species, Agrostis stolonifera (L.) and Carex mackenziei (V. Krecz.). Juncus gerardii (Loisel.) and Calamagrostis stricta (Timm, Koeler) are common throughout the whole meadow. Upper parts of the meadow are characterised by Festuca rubra and locally with willow shrubs. Minor variation in topography breaks zoning by creating mosaic like vegetation. Phragmites and Carex mackenziei typically grow in most shallow depressions and Festuca on low hillocks. Since 1986, Tömppä meadow has been managed by mowing, successfully reducing the previously expanding reed beds and areas of willow scrub. The area lies within the mid-boreal coniferous forest zone and is subject to compensatory land-uplift of ca 8 mm per year (Siira 1970). The area is scheduled under the National Conservation Programme of Wetlands in Finland, confirmed by the Council of State in 1982, and it has been included in Project Mar, as a proposed Ramsar wetland of international importance and proposed Natura 2000 site. The LWfG working team of WWF Finland annually carries out observation work at Tömppä during the entire spring staging period of the LWfG as a part of the annual monitoring and research program. Habitat quality Our working hypothesis was that LWfG staging along the Bothnian Bay prefer the most extensive coastal meadows. To test this the 9 meadows used by LWfG during the last 10 yr were compared with 20 randomly sampled meadows throughout the Bothnian Bay coast between Kalajoki, 64 15 N and Kuivaniemi 65 30 N. Latitude and longitude co-ordinates of 20 points were randomised and the nearest meadow to each was included in the sample. The width of the meadow was measured from 1:20 000 National Survey Board maps (1980s edition) between the wooded zone to the landward side and the shoreline. Under conditions prevailing throughout the Bothnian Bay this equates to the geolittoral zone situated between the average sea level and the high water level (Siira 1970). The hydrolittoral zone between the (average) sea level and the lowest water level was roughly measured determining the distance between the shoreline and the dotted line on the maps showing ca minus one metre under the average sea level. The difference in the width of the geolittoral and hydrolittoral zones was T-tested (one tail, unequal variances) between LWfG staging places and the reference meadows. Determining diet The easiest method of studying diet is to determine the contents of the digestive system in dead birds. Sedinger and Raveling (1984) applied this method in Canada 706 ECOGRAPHY 26:6 (2003)

goose Branta canadensis L., Budeau et al. (1991) in white-fronted goose Anser albifrons Scopoli and Sterbetz (1978) in Hungary in the white-fronted goose and even in the LWfG. This method enables identification of plant material freshly consumed but requires destructive sampling i.e. killing the birds. An alternative method useful in threatened species involves study of the diet by determination of fragments of plant epidermal tissues in faecal pellets. This method was originally developed and tested by Stewart (1967), studying mammal grazers in East Africa. Owen (1975) tested the applicability of the method in geese, and it was also discussed in detail by Bhadresa (1986). The faecal method has been widely applied in diet selection studies in e.g. pink-footed Anser brachyrhynchus Baillon, white-fronted and Canada goose, brent Branta bernicla L. and barnacle goose Branta leucopsis Berchstein (Owen 1976, Buchsbaum et al. 1986, Madsen and Mortensen 1987, Prop and Deerenberg 1991, Spilling et al. 1994). We collected 200 droppings in late May after the LWfG had continued their migration northwards. They were dried at room temperature and ground for microscopic examination. In the field it was not possible to determine droppings from different LWfG individuals, thus the sample is more or less a random sample of all LWfG droppings at Tömppä meadow. The sample potentially contained droppings of up to 31 LWfG individuals. In addition to LWfG, the meadow is used by greylag geese Anser anser anser L. and to a lesser extent bean geese Anser fabalis fabalis Latham, but the faecal pellets of the LWfG can be easily discriminated because they are ca 30% thinner and shorter than the droppings of the larger species. To enable identification of different plant species in droppings, we collected specimens of all the 48 vascular plant species available at Tömppä meadow after the LWfG had left the area in spring 1993 and in some other parts of the Hailuoto Island in November 1994, when the plants were practically in the same condition as in spring. Epidermal tissues which are the most effective to differentiate between plant species of different parts of the plants were prepared for reference samples using the method of Metcalfe (1960) without the use of staining. Reference samples were photographed using a microscope camera combination (Fig. 1). Plant epidermal fragments in goose droppings were compared with the reference samples. We sampled 100 of the 200 dried droppings and ground them manually to avoid excessive destruction of tissues. Then we mixed the sample carefully and one tenth sub-sample was diluted in 70% ethanol and left for some hours. After incubation, a 1/5 sub-sample was transferred to glycerine on a slide. Following Owen s (1975) instructions the sample was spread evenly under the cover glass. Using the point-quadrat method (Owen 1975, Bhadresa 1986) we identified plant fragments in droppings and logged these according to species or taxon and the type of tissue concerned. We sampled the points determining their location as pairs of numbers taken from the random number table and using these figures as co-ordinates (x,y) of the movable objective table of the microscope. Leaves and leaf sheath epidermis tissues were differentiated as upper and lower epidermis, enabling identification of one or both in any one fragment. Where only one epidermis layer was detected, this was noted with details of which side was involved. In all, 2000 observations were made, 20 from each dropping. Mechanical damage may cause leaf fragments to have both or only one of the lower and the upper epidermis present. Thus overestimation of leaf fragments of less robust species might occur if all epidermal fragments were simply summed and compared. Some structural features in the epidermal cells of a number of species complicate identification. For example Festuca rubra and Calamagrostis stricta can only be separated by the lower epidermis. Epidermis on both leaf sides is very similar in Agrostis-species (A. stolonifera, A. gigantea Roth) and they were totally inseparable from each other, in addition, if only one epidermal side was present, separation from upper epidermis of many other Poaceae was impossible. Variation of robustness to physical damage and the close similarity in upper epidermal patterns of most grasses, resulted in calculations of relative proportions of grass items being based upon the number of fragments of lower epidermis only. In addition we corrected the figures of plants available and used (Table 2) to get data sets compatible. Some diet classes had to be removed and some species in the meadow data had to be clumped (e.g. sedges to form a collective class of Carex spp.). Categories Eleocharis/Juncus, monocotyledonous and unidentified plants were deleted. Poaceae items with only the upper epidermis, not recognizable at the species level, included Agrostis leaf fragments, which had only one epidermal side. The number of Agrostis items lost into Poaceae was estimated using Owen s (1975) counts, which showed that in white-fronted goose droppings the proportion of Agrostis items holding both layers of epidermis in our case 9 items was 43%. Calamagrostis/Festuca lower epidermis category was divided between the taxa using the proportion gained according to confirmed observations of these species. These corrections were made to every dropping and every vegetation plot. Availability Availability of potential dietary plant species was assessed by estimating percent coverage of all plant species in 169 random 1 m 2 vegetation plots of the ECOGRAPHY 26:6 (2003) 707

Fig. 1. Light microscope photos of epidermal tissues of the reference material: leaf under side epidermis of (a) Festuca rubra ( 227), (b) Phragmites australis ( 231), (c) Calamagrotis stricta ( 221), (d) Triglochin palustris ( 215). meadow area. This was done in June and July when plant identification was easy. The scale used was +, 0.5, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 40, 50, 60, 70, 75, 80, 85, 90, 95, 97, 98, 99, 100%. To assess food selection and preference, we modified the percentage cover values to the percentage of the total plant cover (leaving bare soil and litter away), for each species in each vegetation plot. Preference and diet selection The geese prefer some plant species and may avoid others (e.g. Owen 1976, Sedinger and Raveling 1984, Madsen and Mortensen 1987, Prop and Deerenberg 1991). The preference for a species can be assessed by comparing the proportion of the species in the diet with the proportion of the species available in vegetation as food (Krebs 1989). We measured diet selection by the LWfG using the electivity index ( ) presented by Chesson (1983). It is based on Manly s alpha selection index ( ) that has been widely applicable in diet studies (Krebs 1989). Using Manly s alpha it is possible to rank plants in order according to frequency in the diet. i = r i 1 m n i (r j /n j ) j=1 where i =Manly s selection index for dietary type (e.g. a plant species) i, r i,r j =proportions of dietary type i and j in the diet (i and j=1, 2, 3,, m), n i,n j =proportion of dietary type i and j available, m=number of potential dietary types. Manly s alpha is applicable in situations where the prey (diet plant) population can be assumed not to be significantly depleted by feeding activity (Krebs 1989). In this study we considered that LWfG exploit a very small proportion of the food available, the quantity of which remains stable. To obtain results that are comparable between cases, where the number of available dietary types varies, we converted Manly s alpha to electivity index presented by Chesson (1983): i = m i-1 (m-2) i +1 Chesson s electivity index potentially ranges between 1 and +1. Dietary types having negative values are 708 ECOGRAPHY 26:6 (2003)

Table 1. Comparison in meadow width between lesser white-fronted goose meadows (n=9) and a random sample of Bothnian Bay meadows (n=20). Geolittoral Hydrolittoral mean min max mean min max LWfG meadows 960 340 1300 978 600 2000 Reference meadows 176 10 860 375 10 1500 avoided (or selected at levels below that expected on the basis of availability) and types getting positive values are preferred (Chesson 1983). If the index value is zero, the food item concerned is taken in the same proportion as it is available. Testing the reliability of electivity indices We tested the reliability of electivity indices using the bootstrap method (Dixon 1993, Efron and Tibshirani 1993). The original i -values were resampled without replacement 1000 times, and an empirical p-value for the preference test was achieved by counting the proportion of p-values that were on the other side of the boundary of neutral selection ( =0) than the original i -value. Results LWfG preferred extensive meadows (Table 1). The average width of the meadow geolittoral favoured by LWfG was 960 m compared to 176 m among the reference meadows (T-test, one-tailed, unequal variances, t=3.548, DF=27, p 0.0001). The hydrolittoral zone showed the same pattern (Table 1). This zone, however, reflects more the condition of future meadows resulting from isostatic up-lift than the situation at present. The average width of the hydrolittoral zone at LWfG meadows was 978 m compared to 375 m in the reference points (T-test, one-tailed, unequal variances, t=2.810, DF=27, p 0.01). Table 2 shows the original (n=2000) and final (corrected) (n=1069) results of dietary items and plants available. About 24% of items could not be identified (e.g. non-epidermal fragments, humus colloids or obscure conglomerations). After corrections, the number of observations per dropping declined to 1 19, with the average of 10.7. The results showed that the diet of the LWfG consists nearly exclusively of monocotyledonous plants. Of all 1524 determined fragments, 99.9% were Monocotyledons of which 87.1% were grasses. The corrections did not radically affect proportions: 99.8% for Monocotyledons and 88.7% for grasses. The three most frequently encountered species were Festuca rubra (43.1%), Phragmites australis (30.2%) and Calamagrostis stricta (13.4%). The most commonly used of other Monocotyledons was Juncus gerardii (8.4%). All these species were common on Tömppä meadows and on the coastal meadows in general, and the LWfG is thus not dependent on any rare plant species. The results revealed two new plant taxa in the diet of the LWfG: Phragmites (a grass) and Eleocharis spp. (other monocotoledonous plants). Indeed, P. australis was the most preferred species. Plant species composition in single droppings provides information on diet selection in short intervals and about variation in diet selection. Of 100 analysed droppings, 22 consisted exclusively of P. australis and 18 of Festuca rubra. The rest were multi-species in content. Their distribution according to the main species was (the average dominance of the main species in parentheses) Festuca rubra 30 droppings (67.9%), Calamagrostis stricta 13 (65.2%), Phragmites 7 (81.2%), Juncus gerardii 6 (54.8%), Triglochin palustris (L.) 2 (61.9%) and Agrostis sp. 1 (66.7%). Table 2. Diet used by the lesser white-fronted goose original and revised data and the diet plants available on Tömppä meadow of the isle of Hailuoto, W Finland. Figures are based on 100 droppings and 169 vegetation sample plots of 1 m 2. Original res. used (%) corrected Available (%) Poaceae Festuca rubra 17.3 43.1 9.2 Calamagrostis stricta 4.2 13.4 19.9 Festuca/Calamagrostis 12.8 Phragmites australis 18.9 30.2 3.9 Agrostis sp. 0.4 1.9 11.1 Poaceae spp. 12.6 Rushes and Cyperaceae Juncus gerardii 4.5 8.4 17.4 Eleocharis sp. 0.05 0.07 8.4 Eleocharis/Juncus 0.05 Carex sp. 0.5 1.3 11.8 Eriophorum angustifolium 0.6 Other Monocotyledonae Triglochin palustris 0.3 1.2 0.2 Triglochin maritima 0.2 Potamogeton sp. 0.1 Monocotyledonae 4.3 Dicotyledonae 0.1 0.2 17.2 Unidentified 23.8 Total (%) 100 100 100 Observations 2000 1069 ECOGRAPHY 26:6 (2003) 709

The results suggest that the LWfG concentrated on feeding on Festuca rubra and Phragmites. Where these two species were abundant locally, the LWfG have not consumed many other species. In multi-species droppings the proportion of the main species was the highest, 81.2%, in Phragmites dominated droppings. The LWfG rarely concentrated on species other than these two and almost never fed exclusively on species other than Festuca or Phragmites. LWfG preferred in descending order Phragmites australis ( =0.73), Triglochin palustris ( =0.7) and Festuca rubra ( =0.52). Calamagrostis stricta ( = 0.47), Juncus gerardii ( = 0.59), Agrostis sp. ( = 0.83), Carex sp. ( = 0.89), Eleocharis sp. ( = 0.99) and Dicotyledons ( = 0.99) were not selected. The LWfG did not exploit monocotyledonous species Triglochin maritima (L.), Potamogeton spp., Eriophorum angustifolium (Honckeny), or a long list of Dicotyledons, all of which had an -value of 1. All the p-values except Triglochin palustris (p= 0.125) were statistically significant (Fig. 2). The great variance of the electivity index in this species was caused by the fact that it was rare both in the field and in the diet. Probably it was preferred, but a larger sample is required to confirm this. Discussion Diet of the lesser white-fronted goose Table 3 summarises the results of diet studies of the LWfG. Our study confirms earlier observations from Hailuoto and nearby areas (Markkola 1992, Markkola et al. 1993) that the spring diet of LWfG consists almost entirely of Monocotyledons. Also in the important staging area at Porsangerfjord, northern Norway, a grass, Puccinellia phryganodes, is normally the main diet. In spring 1996 when the grass meadows were covered by ice until very late in the spring LWfG used a dicotyledonous species Hippuris tetraphylla that was practically the only forage available (Aarvak et al. 1996). Both of these species also grow sparsely at Tömppä meadow but were not found in the diet. Cultivated species were not observed in the diet of LWfG on Hailuoto. Most of the few cases when we have seen LWfG feeding in fields and eating mainly Phleum pratense L. grass were under retarded spring conditions when the meadows were covered by ice (Markkola 1992, 2001). The Hungarian steppe, puszta, has been the westernmost highly important autumn and spring stop-over for the LWfG during the 20th century, in 1950s still hosting up to 50 000 individuals per season (Sterbetz 1978). Sterbetz (1978, 1990) found Festuca pseudo ina Hackel ex Wiesb. grass to be the outstanding fall feed of the LWfG. Most F. pseudo ina steppes with a high calcium content have been ploughed, but they still exist in nature reserves like Hotobágy and Kardoskut (Sterbetz 1990). In wintering areas of Azerbaijan and Armenia, LWfG are to some extent grazing in fields on wheat, barley and maize (Zea mays L.) but they are also said to prefer feeding in steppe grasslands where sheep grazing maintains low vegetation growth (Lorentsen et al. 1999). At East Dongting Lake, China (29 10 N, Fig. 2. The distributions of Chesson s electivity indices and the empirical p-values of 1000 bootstrap re-samples: grey rectangles=50% of values, black line inside the rectangle=median, segment of line=holding 80% of observations, circles=extreme 20% of observations. 710 ECOGRAPHY 26:6 (2003)

Table 3. Known dietary species of the lesser white-fronted goose. Preferred species have a rank number (1) after their Latin name. In most short notes about the diet of the LWfG there is nothing mentioned about the preference and the species are in a random order. *=cited by Cramp and Simmons (1977). Source dietary plants area season this study (the order is according to grasses: Phragmites australis (1), Isle of Hailuoto, Bothnian Bay, spring the electivity index) Festuca rubra (3), Calamagrostis Finland Markkola 1992 stricta (4), Agrostis sp. (6) other Monocot.: Triglochin palustris (2), Juncus gerardii (5), Carex sp. (7), Eleocharis sp. (8) Dicot.: Dicotyledonae sp. (9) grasses: Agrostis stolonifera, Isle of Hailuoto, Bothnian Bay, Festuca rubra, Calamagrostis stricta other Monocot.: Juncus gerardii Finland Markkola 1992, Markkola et al. grasses: Phleum pratense (coastal Bay of Liminganlahti, Bothnian 1993 fields) Bay, Finland Aarvak et al. 1996 other Monocot.: Carex halophila, Carex mackenziei, C. paleacea grasses: Puccinellia phryganodes (1) Porsangerfjord (Valdak), Norway Morozov 1988 Dioct.: Hippuris tetraphylla grasses: Arctophila ful a N part of the Ural Mountains, summer Bolshezemelskaja Tundra Lorentsen and Spjøtvoll 1990 other Monoct.: Carex bigelowii, Nordland, Norway (Scandinavian Carex nigra ssp. nigra, C. nigra Mountains) Nettlebladt 1992 ssp. juncella, Eriophorum scheuzeri, E. angustifolium, Trichophorum cespitosum Dicot.: Polygonum i iparum, Leontodon autumnalis, Taraxacum sp. horsetails: Equisetum palustre other Monocot.: Carex aquatilis, Nordland, Norway Eriophorum angustifolium, Juncus arcticus, Luzula multiflora Dicot.: Petasites frigidus, Polygonum i iparum, Salix lanata horsetails: Equisetum palustre Markkola 1992, Markkola et al. other Monocot.: Carex rostrata, C. Lapland, Finland 1998a chordorrhiza Dicot.: Empetrum nigrum berries *Scott ym. Dicot.: Salix herbacea Lapland/Finnmark *Zharkova and Borzhonov other Monocot.: Eriophorum Taimyr, Russia angustifolium, E. scheuzeri Aarvak et al. 1996 grasses: Festuca rubra, Puccinellia Porsangerfjord (Valdak), Norway autumn phryganodes, Agrostis stolonifera, Elymus arenarius other Monocot.: Eleocharis uniglumis, Juncus gerardii Tolvanen 1998 grasses: Puccinellia phryganodes Kanin Peninsula, Russia other Monocot.: Carex subspathacea Dicot.: Hippuris tetraphylla Tolvanen et al. 1998 Dicot.: Empetrum nigrum berries Varangerfjord, Norway Markkola et al. 1998b Dicot.: Empetrum nigrum berries Isle of Hailuoto, Finland Sterbetz 1978 (1990 in text) grasses: Festuca pseudo ina (1), Carpathian Basin, Hungary Markkola et al. 2000 Gramineae sp., Triticum ulgare, Poa sp. Dicot.: Chenopodium sp., Achillea sp., Sinapis sp., Eryngium sp. Dicot.: Rorippa sp. China winter ECOGRAPHY 26:6 (2003) 711

113 50 E), where 50% of the known world population of the LWfG are currently wintering, LWfG mostly used grasslands (88%), where the dominant plants were sedges and grasses (Markkola et al. 2000). The gizzard of a poisoned 2nd calendar-year male LWfG, however, contained only Rorippa sp., a dicotyledon (Markkola et al. 2000). In summer, the fundamental difference compared with migration and wintering is the molting and brood rearing period, when a LWfG family does not fly for 6 7 weeks. One could expect that summer diet is not as limited as in other parts of the year. The 18 species listed for summer (Fennoscandia, Table 3) compared with the 9 taxa of this study or all 14 species listed for spring (Finland Norway, Table 3) support this idea. The summer diet also differs from the normal grass dominated pattern. Despite the huge geographical scale concerned, there is a lot in common in the diet of the LWfG and the other geese. However, in staging areas, LWfG prefer natural habitats more than other geese. In Hungary, LWfG prefer natural steppe whilst white-fronts and bean geese mostly graze in fields (Sterbetz 1978, 1990). Along Bothnian Bay coasts in Finland, LWfG mostly use coastal meadows, where the taiga bean geese use fields (Markkola 1992, 2001, Markkola et al. 1998a). This dependence on specific places supporting special habitats may make the LWfG vulnerable to continued habitat loss and illegal hunting in the few remaining concentrations. Factors affecting diet selection Based on the results of this study and existing knowledge, the LWfG is not strictly limited to a few particular plant species. Probably the most important factor leading to a limited list of dietary species is its need for a special feeding habitat, which according to our results is an extensive coastal meadow, at least tens of hectares wide and 300 m broad. The LWfG is, at least in Bothnian Bay staging areas, more shy than other geese (Markkola et al. 1998b). Disturbance from people and vehicles is less frequent on the remote coastal meadows than in agricultural areas. Also Sterbetz (1990) stressed the lack of disturbance in large puszta areas as a reason for the LWfG to use this habitat in addition to the morphological feature of the LWfG, which has a short bill adapted to intake of short-growth plants. Within suitable habitats, the quality of different plants is probably the main factor leading to diet selection by the LWfG. The quality of food varies in different components such as the protein content, soluble carbohydrates, energy and digestibility. We did not carry out chemical analyses and the data in the literature are heterogeneous (Thomas and Prevett 1980, Sedinger and Raveling 1984, Budeau et al. 1991). Therefore it was difficult to conclude the importance of the chemical composition in diet selection. However, some patterns were found: the content of soluble carbohydrates, but not that of protein, ranked the species in the same order as their favour by LWfG. The rank in fibre content, as expected was near the opposite to the rank in LWfG preference. Management of lesser white-fronted goose staging meadows The traditional use of coastal meadows of the Bothnian Bay was mowing in mid-summer and grazing of cattle in late summer, a practice, which ceased gradually in the 1950s. Since 1986 the Tömppä meadow has been managed again by mowing, thanks to the local hunting association which has tried to improve conditions for waterfowl and hunting and conservationists who have tried to improve conditions for threatened bird and plant species. Mowing has improved the conditions for staging LWfG, breeding birds such as southern dunlins (Calidris alpina schinzii Brehm) and for rare plant species like the arctic salt-grass (Puccinellia phryganodes) and the Siberian primrose (Primula nutans Georgi) (Markkola and Merilä 1996). Encroachment by reed beds and willow bushes that have invaded the open meadows has been reduced. The draft management plan for the area (Markkola and Merilä 1996) suggested mowing and grazing according to the traditional management scheme. It is, however, more probable that the new management method will be continuous grazing in June August, as natural pastures are effectively promoted in the EU through agri-environment scheme support. Grazing can even, more effective than mowing, improve conditions for plants, which are preferred by the LWfG. The common reed Phragmites australis can have an adverse effect when it encroaches on open coastal meadows but at the same time it was the most preferred dietary species of the LWfG. LWfG never fed in dense and tall reed stands, but searched out short and scattered reed shoots. In an ideal situation Phragmites will survive sparsely among lower vegetation. To ensure the survival of sparse Phragmites stands, the meadow could be divided into management segments, of which each could be cut every second year or in two years out of three. If grazing begins, maintaining of Phragmites requires moderate grazing pressure. The negative effects of grazing on reed will probably be balanced by expansion of Triglochin palustris and Puccinellia phryganodes, which are capable of colonising the bare substrate created by cattle (Kauppi 1967). Where pasture can be divided into two three segments, one two could be grazed and one two left untouched annually to favour also breeding birds, as trampling of nests is avoided in untouched segments (de Jong 1977). 712 ECOGRAPHY 26:6 (2003)

One important point in management of the LWfG meadows of Hailuoto is the need for a hunting ban in this area, which is a candidate site for inclusion in the Natura 2000 network of the EU. At the moment, wildfowl shooting occurs during the main autumn migration period through the area, which is also a potential autumn staging place for LWfG. Acknowledgements This study is a part of the study and conservation work carried out by the Lesser White-fronted Goose working group of WWF Finland and Finnish conservation authorities, the Nordic Lesser White-fronted Goose Project (including NOF/BirdLife Norway) and the Lesser White-fronted Goose Task Force of Wetlands International. The Finnish Academy, Maj and Tor Nessling Foundation and NorNet/Focus area Environment of the Univ. of Oulu have supported this study financially. We thank Tony Fox, Seppo Haapala, Jari Heikkinen, Satu Huttunen, Anja Hohtola, Marko Hyvärinen, Kari Koivula, Kimmo Lahti, Eino Merilä, Helvi Mikkola, Hanna-Liisa Suvilampi, Robert Thomson and Mika Ylitervo. References Aarvak, T., Øien, I. J. and Nagy, S. 1996. The Lesser Whitefronted Goose Monitoring Programme. Annual report 1996. NOF Rapportserie, Rep. No. 7-1996, Norwegian Ornithological Society, Klæbu. Aarvak, T. et al. 1999. Two pieces of the spring migration puzzle of the Fennoscandian lesser white-fronted goose population in place. In: Tolvanen, P., Øien, I. J. and Ruokolainen, K. (eds), Fennoscandian Lesser Whitefronted Goose Conservation Project. Annual report 1998. WWF Finland Rep. 10 and Norwegian Ornithological Society, NOF Rapportserie Rep. No. 1-1999, pp. 27 30. Aarvak, T. et al. 2001. Introduction. In: Tolvanen, P., Øien, I. J. and Ruokolainen, K. (eds), Fennoscandian Lesser White-fronted Goose Conservation Project. Annual report 2000. WWF Finland Rep. 13, 1998 No. 10 and Norwegian Ornithological Society, NOF Rapportserie Rep. No. 1-2001, pp. 5 8. Bhadresa, R. 1986. Faecal analysis and exclosure studies. In: Moore, P. D. and Chapman, S. B. (eds), Methods of plant ecology. Blackwell, pp. 437 524. Buchsbaum, R., Wilson, J. and Valiela, I. 1986. Digestability of plant constituents by Canada geese and Atlantic brant. Ecology 67: 386 393. Budeau, D. A., Ratti, J. T. and Ely, G. R. 1991. Energy dynamics, foraging ecology, and behavior of prenesting greater white-fronted geese. J. Wildl. Manage. 55: 556 563. Chesson, J. 1983. The estimation and analysis of preference and its relationship to foraging models. Ecology 64: 1297 1304. Cramp, S. and Simmons, K. E. L. (eds) 1977. The birds of the western Palearctic, 1. Oxford Univ. Press. de Jong, H. 1977. Experience with the man-made meadow bird reserve Kievitslanden in Flevoland (the Netherlands). Biol. Conserv. 12: 13 31. Dixon, P. M. 1993. The bootstrap and the jackknife: describing the precision of ecological indices. In: Scheiner, S. M. and Gurevitch, J. (eds), Design and analysis of ecological experiments. Chapman and Hall, pp. 290 318. Ebbinge, B. et al. 1982. The importance of spring staging areas for arctic breeding geese, wintering in western Europe. Aquila 89: 249 258. Efron, B. and Tibshirani, R. J. 1993. An introduction to the bootstrap. Monographs on statistics and applied probability 57. Chapman and Hall. Iwabuchi, S. et al. 1997. The short report on survey of lesser white-fronted geese at Donting lakes in China. Bull. Sci. Mus. 7: 103 105. Kauppi, M. 1967. U ber den Einfluss det Beweidung auf die Vegetation der Uferwiesen an der Bucht Liminganlahti im Nordteil des Bottnischen Meerbusens. Aquilo Ser. Bot. 6: 347 368. Krebs, C. J. 1989. Ecological methodology. Harper Collins. Lorentsen, S.-H. and Spjøtvoll, Ø. 1990. Notes on the food choice of breeding lesser white-fronted goose Anser erythropus. Fauna Norvegica Ser. C, Cinclus 13: 87 88. Lorentsen, S.-H., Øien, I. J. and Aarvak, T. 1998. Migration of Fennoscandian lesser white-fronted geese Anser erythropus mapped by satellite telemetry. Biol. Conserv. 84: 47 52. Lorentsen, S.-H. et al. 1999. Lesser white-fronted goose Anser erythropus. In: Madsen, J., Fox, T. and Cracknell, J. (eds), Goose populations of the Western Palearctic. Wetlands International, Cambridge, pp. 144 161. Madsen, J. 1996. International action plan for the lesser white-fronted goose (Anser erythropus). In: Heredia, B., Rose, L. and Painter, M. (eds), Globally threatened birds in Europe. Action plans. Council of Europe Publishing, BirdLife International, pp. 67 78. Madsen, J. and Mortensen, C. E. 1987. Habitat exploitation and interspecific competition of moulting geese in east Greenland. Ibis 129: 25 44. Markkola, J. 1992. The lesser white-fronted goose. In: Threatened animals of the world. Finland. Weilin+Göös, Vantaa, pp. 96 99, in Finnish. Markkola, J. 2001. Spring staging of lesser white-fronted geese on the Finnish Bothnian Bay coast in 2000. In: Tolvanen, P., Øien, I. J. and Ruokolainen, K. (eds), Fennoscandian Lesser White-fronted Goose Conservation Project. Annual report 2000. WWF Finland Rep. 13 and Norwegian Ornithological Society, NOF Rapportserie Rep. No. 1-2001, pp. 12 16. Markkola, J. and Merilä, E. 1996. The preliminary management plan for the protected area of Iso Matala-Härkäsäikkä on the isle of Hailuoto, Bothnian Bay, Finland. Pohjois-Pohjanmaan ympäristökeskus (Regional Environment Centre of Northern Ostrobothnia). Mare Botnicum Ay. Oulu, in Finnish with English summary. Markkola, J. et al. 1993. Liminganlahti. Pohjoinen. Kirjapaino Osakeyhtiö Kaleva, Oulu. Markkola, J., Bianki, V. and Zimin, V. 1998a. The lesser white-fronted goose (Anser erythropus). In: Kotiranta, H. et al. (eds.), Red Data Book of east Fennoscandia. Ministry of the Environment, Finnish Environment Inst. and Finnish Museum of Natural History, Helsinki, pp. 220 224. Markkola, J., Ohtonen, A. and Karvonen, R. 1998b. Spring staging areas of the lesser white-fronted goose Anser erythropus on Bothnian Bay coast: features of spring migration in 1997. In: Tolvanen, P. et al. (eds), Finnish Lesser White-fronted Goose Conservation project. Annual report 1997. WWF Finland Rep. No 9, pp. 12 17. Markkola, J. et al. 2000. Lesser white-fronted goose survey at East Dongting and Poyang lakes in China, February 1999. In: Tolvanen, P., Øien, I. J. and Ruokolainen, K. (eds), Fennoscandian Lesser White-fronted Goose Conservation Project. Annual report 1999. WWF Finland Rep. 12 and Norwegian Ornithological Society, NOF Rapportserie Rep. No. 1-2000, pp. 9 15. Metcalfe, C. R. 1960. Anatomy of the monocotyledons. I Gramineae. Clarendon Press. Morozov, V. 1988. The lesser white-fronted goose in eastern Bolshezemelskaja Tundra and Polar Urals. Resources of rare animals in RSFSR, their conservation and use. Moscow, pp. 71 77, in Russian. Nettlebladt, D. N. 1992. Rapport fra registrering av vegetasjon i dverggåsområden i Bjorådalen, Rana. Norsk Ornitologisk Forening (NOF/BirdLife). ECOGRAPHY 26:6 (2003) 713

Nilsson, L. 1979. Variation in the production of young of swans wintering in Sweden. Wildfowl 30: 129 134. Norderhaug, A. and Norderhaug, M. 1984. Status of the lesser white-fronted goose, Anser erythropus, in Fennoscandia. Swedish Wildl. Res. 13: 171 185. Owen, M. 1975. An assessment of faecal analysis technique in waterfowl feeding studies. J. Wildl. Manage. 39: 271 279. Owen, M. 1976. The selection of winterfood by white-fronted geese. J. Appl. Ecol. 13: 715 729. Prop, J. and Deerenberg, C. 1991. Spring staging in brent geese Branta bernicla: feeding constraints and the impact of diet on the accumulation of body reserves. Oecologia 87: 19 28. Ruokonen, M. 2001. Phylogeography and conservation genetics of the lesser white-fronted goose (Anser erythropus). Ph.D. thesis, Acta Universitatis Ouluensis A 360, Oulu. Sedinger, J. S. and Raveling, D. G. 1984. Dietary selectivity in relation to availability and quality of food for goslings of cackling geese. Auk 101: 295 306. Siira, J. 1970. Studies in the ecology of the sea-shore meadows of the Bothnian Bay with special reference to the Liminkaarea. Aquilo Ser. Bot. 9: 1 109. Soulé, M. (ed.) 1986. Conservation biology. The science of scarcity and diversity. Sinauer. Spilling, E., Bergmann, H.-H. and Stock, M. 1994. Diet of dark-bellied brent geese (Branta b. bernicla) in the Piassina delta, Taimyr, Siberia. Progress report. IWRB Goose Res. Group Bull. 5: 18 25. Sterbetz, I. 1978. The feeding ecology of Anser albifrons, Anser erythropus and Anser fabalis in Hungary. IWRB Bull. 45: 9 16. Sterbetz, I. 1990. Variations in the habitat of the lesser whitefronted goose (Anser erythropus L., 1758) in Hungary. Aquila 96 97: 11 17. Stewart, D. R. M. 1967. Analysis of plant epidermis in faeces: a technique for studying the food preferences of grazing herbivores. J. Appl. Ecol. 4: 83 111. Thomas, V. G. and Prevett, J. P. 1980. The nutritional value of arrow-grasses to geese at James Bay. J. Wildl. Manage. 44: 830 836. Tolvanen, P. 1998. Lesser white-fronted goose Anser erythropus expedition to the Kanin Peninsula in 26 August 12 September, 1996, and the establishment of the Shoininsky Reserve. In: Tolvanen, P. et al. (eds), Finnish Lesser White-fronted Goose Conservation Project. Annual report 1997. WWF Finland Rep. No. 9, pp. 33 35. Tolvanen, P. and Markkola, J. 1998. Introduction. In: Tolvanen, P. et al. (eds), Finnish Lesser White-fronted Goose Conservation Project. Annual report 1997. WWF Finland Rep. No. 9, pp. 5 6. Tolvanen, P., Pynnönen, P. and Ruokolainen, K. 1998. Monitoring of lesser white-fronted goose Anser erythropus on Skjåholmen (Varangerfjord, Finnmark, Norway) in 1995 1997. In: Tolvanen, P. et al. (eds), Finnish Lesser Whitefronted Goose Conservation Project. Annual report 1997. WWF Finland Rep. No. 9, pp. 30 32. 714 ECOGRAPHY 26:6 (2003)