Numbers of wintering waterbirds in the Czech Republic: long-term and spatial-scale approaches to assess population size

Abstract

Capsule Two nationwide monitoring schemes, the International Waterbird Census (IWC) and Atlas of Wintering Birds (AWB), provide comparable estimates of national population sizes of wintering waterbirds. Differences between these estimates for several species can be explained by differences in their migratory strategy.

Aims We tested whether species-specific variables, i.e. the total numbers, distribution and migration strategy affected the accuracy of wintering population estimates provided by two different monitoring schemes, the IWC and AWB.

Methods Estimates of the numbers of 81 waterbird species are compared, using data from IWC (mid-January) and AWB (December, January and February) in the Czech Republic between 1982 and 1985. Log-linear Poisson regression analysis was used to estimate missing IWC data from long-term IWC data series (1966–2013) using Trends and Indices for Monitoring data software. A series of generalized linear models was estimated to analyse the effect of species-specific variables on the presence of overlap between IWC and AWB national population estimates.

Results The range of numbers recorded by IWC data overlap the range of AWB numbers in 22 of the 53 investigated species recorded by both monitoring schemes. Numbers of most other species based on the IWC counts were lower when compared with those generated from AWB data and included several widespread species, e.g. White-tailed Eagle Haliaeetus albicilla and White-throated Dipper Cinclus cinclus. In contrast, population size estimates based on IWC counts were higher than those generated from AWB estimates for one species, the Ferruginous Duck Aythya nyroca. The differences between IWC and AWB estimates were significantly higher in late-departing and early-arriving migrants. There was no effect of species distribution and population size.

Conclusions Based on a single-country data sample, we support the relevance of the IWC Census methodology to produce estimates of wintering waterbird numbers and to calculate 1% thresholds for waterbird species, which we use to indicate nationally important wetland sites for future conservation and protection in the Czech Republic.

Waterbirds are among the most well-studied of animal groups, and several monitoring methods have been developed involving large numbers of volunteers in several long-term annual programmes (Bibby et al. 1992, Wetlands International 2006). Wintering waterbirds often occur in spectacular concentrations while using a limited network of sites, the proper identification and protection of which has proved to be an effective conservation mechanism for many waterbird species (Jackson et al. 2004b). Moreover, waterbirds are recognized as indicators of the richness and diversity of wetland ecosystems (Bibby et al. 1992, Boere et al. 2006).

Up to now, the majority of important European wintering sites for most waterbird species have been found in the coastal areas of northwest Europe, the Baltic Sea and in the Mediterranean region (Gilissen et al. 2002, Kershaw & Cranswick 2003a, Skov 2011, Van Roomen et al. 2012, Lehikoinen et al. 2013). The importance of central Europe as a wintering region for waterbirds has been increasing in the recent decades (Fox et al. 2010, Musil et al. 2011). In particular, lakes, unfrozen reservoirs and running waters have attracted around 200 000 waterbirds every winter in central Europe (Gilissen et al. 2002, Keller 2011). The relatively mild climate can provide feeding opportunities throughout the wintering period especially when the Baltic coast becomes frozen in severe winters (Švažas et al. 2001, Nilsson 2008). There has thus been a long-term (1966–2008) increase in numbers counted in 20 of the 26 most abundant wintering waterbird species in the Czech Republic and an expansion in range in 16 of these species that broadly correlates with the changes in numbers (Musil et al. 2011).

Based on these changes in numbers of wintering waterbirds, as well as their distribution, we can expect a redistribution of waterbirds in their European wintering ranges (Lehikoinen et al. 2013). Estimation of population size is therefore essential for the designation and conservation of important staging and wintering sites. Among the criteria for designation of wetlands of international importance (Ramsar sites), the 1% threshold (Criterion No. 6) is frequently used: a wetland should be considered internationally important if it regularly supports 1% of the individuals in a population of one species or subspecies of waterbird (Wetlands International 2006). This national 1% threshold is regularly used in many countries, such as Germany, Ireland, the UK and Slovakia (Wahl & Sudfeldt 2005, Crowe et al. 2008, Calbrade et al. 2010, Slabeyová et al. 2011). In this context, knowledge of population size is fundamental to the application of conservation planning and action (Jackson et al. 2004a, Sutherland et al. 2004), either using the 1% flyway or national threshold value (Wetlands International 2006).

The relevant data for the assessment of wintering numbers are provided by nationwide monitoring schemes. The International Waterbird Census (IWC) monitors wintering waterbirds on a global level on the basis that counts are coordinated over a relatively short period of time in January across the respective flyways (Wetlands International 2006). In this case, monitoring is able to indicate both population sizes and their changes over the long-term (Musil et al. 2011). Bird atlas studies (Bibby et al. 1992) are also used to describe distribution and numbers using sample units over a regular grid. Atlases of wintering birds have been published in several European countries including the Czech Republic (Lack 1986, SOVON 1987, Bejcˇek et al. 1995, Yeatman-Berthelot & Jarry 1995). Atlas studies are not confined to wetlands, rather they are usually based on coverage of entire regions, countries or geographical units.

Here, we compare data from two such contrasting nationwide monitoring schemes, the IWC and the Atlas of Wintering Birds (AWB). In the light of different methodological approaches, this study tests the use of long-term IWC data using the more comprehensive AWB data. We expect that patterns in species distribution and/or population size may result in differences between the IWC and AWB estimates. If there is a density-dependent bias, the probability of different IWC or AWB estimates will be correlated with population size or with the number of sites occupied by individual species. Bias may also arise because of species dispersion, i.e. it may be lower in widely represented species (Bibby et al. 1992). Moreover, the migratory strategy of waterbird species may result in differences between IWC and AWB estimates, because of the different scale of seasonal coverage of the IWC (mid-January) and AWB (December–January–February) counting schemes. We therefore try to assess if any differences between IWC and AWB estimates of wintering waterbird population sizes can be explained by three species-specific characteristics: their distribution, population size and migratory strategy. Finally, our comparisons also allow us to provide actual (2009–2013) estimates of national wintering population sizes of waterbird species and so to identify the 1% national threshold value for individual species in the Czech Republic.

METHODS

In total, 81 species of birds that are ecologically dependent upon wetlands (Wetlands International 2006) were included in this study; these include divers, grebes, cormorants, herons, bitterns, storks, swans, geese, ducks, cranes, rallids, waders, gulls, terns, White-tailed Eagle Haliaeetus albicilla, wagtails, White-throated Dipper Cinclus cinclus and Bearded Tit Panurus biarmicus (Gilisen et al. 2002, Musil et al. 2011). The three gull species Herring Gull Larus argentatus, Caspian Gull Larus cachinnans and Yellow-legged Gull Larus michahellis were termed ‘large gulls’, and hereafter are treated as one single species, in accordance with the taxonomic situation valid at the beginning of the study period (Rose 1995, Musil et al. 2011, Wetlands International 2013).

Numbers of individuals of waterbird species are counted annually worldwide for the IWC, which is carried out each winter in mid-January. Counts are conducted on predetermined dates with the aim to maximize synchrony. Within the Czech Republic, the IWC counts have been carried out annually at between 48 and 639 wetland sites in January each year between 1966 and 2013. A total of 973 sites (Fig. 1) have been surveyed overall including all wetland sites counted in at least two winters between 1966 and 2013 (Musil et al. 2011). The regional coverage of the Czech Republic (Fig. 1), as well as the proportion of sites classified as running and standing waters, has remained similar in all counts from 1966 to 2013 (Fiala 1980, Musil & Musilová 2010). For this study, data from all 973 wetland sites investigated in the Czech Republic between 1966 and 2013 (Fig. 1) were used. Among those, 248 wetland sites were counted in detail between 1982 and 1985. A countrywide monitoring of the distribution of wintering birds was carried out by the AWB in the months of December, January and February between 1982 and 1985 (Bejcˇek et al. 1995), using grid ‘squares’ that were 12 × 11.1 km in size (which equates with 10 minutes longitude by 10 minutes latitude). Data were obtained from 595 out of a possible total of 611 squares (Fig. 1). The fieldwork methods used and data analysis followed that of the published AWB in Britain and Ireland (Lack 1986). Maps of individual species were based on the highest count of each species seen in a square on any one day during the survey period (Lack 1986). Two types of records were accepted, those from regular visits to squares as well as any supplementary (casual) records of individual species (Lack 1986). The numbers of individual species in a quadrat were estimated in the following categories: 1–10, 11–100, 101–1000, 1001–10 000, 10 001 and more individuals (Bejcˇek et al. 1995).

Figure 1. Distribution of wetland sites investigated in 1982–85 (solid circles) and investigated in other years between 1966 and 2013 (open circles) in the Czech Republic. The grid of squares represents squares used by AWB (12 km × 11.1 km in size). Grey squares were not investigated.

Log-linear Poisson regression analysis was used to estimate missing 1982–1985 IWC data from long-term IWC data series (1966–2013) using Trends and Indices for Monitoring data (TRIM) software (Statistic Netherlands version 3.52, Pannekoek & Van Strien 2005). Missing data resulted from incomplete coverage due to limited availability of volunteers in some seasons. Serial correlations between annual numbers and over-dispersion in the data were also taken into account. The models used included change points to allow for changes in the slope parameters at some points in the time series (Pannekoek & Van Strien 2005, Fouque et al. 2007, 2009). ‘Time Totals’ values (hereafter used as Time Totals) of IWC data (i.e. the actual count values plus the numbers of birds estimated by the TRIM software) for all 973 sites included in the analysis were used to generate total estimates of the range of numbers of the waterbird species wintering in the Czech Republic. These Time Totals were then rounded according to the guidelines used in Waterfowl Population Estimates (Wetlands International 2006, 2013). We use the range (min–max) of population estimates due to the effect of between-year variation in numbers because of variable climatic conditions (Musil et al. 2008).

For the AWB, estimates of wintering numbers (i.e. the min–max range of the wintering population size) of individual bird species were calculated by summing the numbers of birds recorded on individual squares (for detailed methodology see Bejcˇek et al. 1995). Finally, population sizes were also estimated using recent IWC data (in a similar manner to the calculations used to generate estimates for the 1982–1985 period) between 2009 and 2013, and 1% national thresholds were calculated for all waterbird species in the Czech Republic whose mean wintering population size exceeded 100 individuals (Kershaw & Cranswick 2003a).

The aim of the statistical analyses was to assess the impact of species-specific characteristics on any differences between the IWC and AWB national population estimates. We worked with three alternative definitions of an overlap between the two counts. The first definition was simply whether overlap occurred at all (e.g. a binary variable, yes/no). The other two definitions, hence referred to as overlap proportion (IWC) and overlap proportion (AWB), take the ratio of the overlap length to the spread of the IWC and AWB ranges, respectively: see Fig. 2 for an illustration of the calculations. We considered the following continuous species-specific characteristics: the number of occupied sites (IWC) or squares (AWB) (i.e. a measure of a species' distribution) and median population estimates (i.e. a measure of population size). Both these continuous variables can be specified based either on IWC or AWB records; we decided to try both and run two separate analyses in parallel, one with IWC versions of the distribution and population size variables and one with the AWB counterparts. In both cases the variables were entered into the regressions in log form. Moreover, we included a qualitative indicator of a species' migratory strategy. We distinguished two types of migratory strategy: (i) late-departing/early-arriving migrants (Newton 2008), coded as 1 and (ii) others (resident species), coded as 0. We used the term ‘late-departing migrants’ for species with prolonged migration up to December and ‘early-arriving migrants’ for species which spring migration can start early, i.e. in February. Their numbers can be higher either at the beginning (in December) and/or at the end (in February) of the winter (i.e. ABW monitoring period) than in mid-January (i.e. IWC census term). These species represent mostly those where breeding population size exceeded wintering population or those where peak numbers are usually reached in autumn and spring migration, which can include also December and/or February (Hudec 1994, Hudec et al. 1999, Musil et al. 2001). For the binary overlap variable, we applied a conventional logistic regression, while for the fractional outcomes we used the quasi-maximum likelihood estimator of Papke & Wooldridge (1996), sometimes dubbed ‘fractional logit’. Both the logistic regression and the fractional logit were then estimated using the generalized linear model (GLM) routine with the logit link function in Stata 13 (StataCorp, College Station, TX, 2013).

Figure 2. Illustration of the calculation of overlap proportion for Goosander.

RESULTS

Among the 81 species, 53 were recorded by both IWC and AWB monitoring schemes, 2 species recorded only by IWC and 26 recorded only by AWB. The range of wintering waterbird population sizes recorded by IWC data overlap the range of the AWB population sizes in 22 of these 53 species (Table 1). The range of AWB numbers exceeded the range of IWC numbers in 30 of those species and the range of IWC numbers exceeded the range of AWB numbers in Ferruginous Duck Aythya nyroca.

Table 1. National population size estimates of wintering waterbirds based on data gathered as part of the IWC and AWB between 1982 and 1985 (mean and max values across the four years).

Common name	Species	IWC	AWB	Overlap	Late-departing/early-arriving migrants
Mute Swan	Cygnus olor	3800–4400	2900–4000	1	0
Tundra Swan	Cygnus columbianus	0	0–3	0	1
Whooper Swan	Cygnus cygnus	6–6	0–15	1	1
Bean Goose	Anser fabalis	960–1800	20 000–40 000	0	1
Pink-footed Goose	Anser brachyrhynchos	0	0–12	0	1
White-fronted Goose	Anser albifrons	15–35	400–800	0	1
Lesser White-fronted Goose	Anser erythropus	0	0–1	0	1
Greylag Goose	Anser anser	80–130	400–900	0	1
Canada Goose	Branta canadensis	0	0–5	0	1
Barnacle Goose	Branta leucopsis	0–1	0	0	1
Common Shelduck	Tadorna tadorna	0	0–1	0	1
Wood Duck	Aix sponsa	0–1	0–1	1	1
Eurasian Wigeon	Anas penelope	40–45	10–50	1	1
Gadwall	Anas strepera	3–9	50–100	0	1
Common Teal	Anas crecca	1100–1400	1200–2500	1	1
Mallard	Anas platyrhynchos	133 000–155 000	100 000–300 000	1	1
Northern Pintail	Anas acuta	9–10	5–14	1	1
Garganey	Anas querquedula	0	0–10	0	1
Northern Shoveler	Anas clypeata	4–7	10–60	0	1
Red-crested Pochard	Netta rufina	1–2	1–15	1	1
Common Pochard	Aythya ferina	1500–1900	3000–7000	0	1
Ferruginous Duck	Aythya nyroca	7–20	1–5	0	1
Tufted Duck	Aythya fuligula	1000–1400	2500–5000	0	1
Greater Scaup	Aythya marila	9–10	5–15	1	0
Common Eider	Sommateria mollissima	0–1	2–10	0	1
Long-tailed Duck	Clangula hyemalis	4–5	2–10	1	0
Common Scoter	Melanitta nigra	0–2	0–5	1	0
Velvet Scoter	Melanitta fusca	3–10	8–20	1	0
Common Goldeneye	Bucephala clangula	590–650	500–1000	1	0
Smew	Mergellus albellus	15–25	30–60	0	1
Red-breasted Merganser	Mergus serrator	6–70	4–8	1	1
Goosander	Mergus merganser	1500–2600	800–2000	1	0
Red-throated Diver	Gavia stellata	0	2–10	0	1
Black-throated Diver	Gavia arctica	3–6	10–30	0	1
Great Northern Diver	Gavia immer	0	0–1	0	1
Little Grebe	Tachybaptus ruficollis	930–990	1000–2000	0	1
Great Crested Grebe	Podiceps cristatus	170–190	400–900	0	1
Red-necked Grebe	Podiceps grisegena	0–2	2–20	0	1
Horned Grebe	Podiceps auritus	0–1	1–15	0	1
Black-necked Grebe	Podiceps nigricollis	0–1	1–10	0	1
Great Cormorant	Phalacrocorax carbo	600–810	500–1500	1	0
Great Bittern	Botaurus stellaris	1–1	0–3	1	1
Night Heron	Nycticorax nycticorax	0	0–1	0	1
Great White Egret	Egretta alba	4–15	50–100	0	1
Grey Heron	Ardea cinerea	1100–1800	3000–6000	0	1
Black Stork	Ciconia nigra	0	0–1	0	1
White Stork	Ciconia ciconia	0	4–10	0	1
Glossy Ibis	Plegadis falcinellus	0	0–1	0	1
Water Rail	Rallus aquaticus	4–4	20–50	0	1
Common Moorhen	Gallinula chloropus	500–570	700–1500	0	1
Common Coot	Fulica atra	16 000–18 000	18 000–36 000	0	1
European Golden Plover	Pluvialis apricaria	0–1	0–3	1	1
Grey Plover	Pluvialis squatarola	0	0–1	0	1
Northern Lapwing	Vanellus vanellus	10–180	800–1600	0	1
Little Stint	Calidris minuta	0–1	0	0	1
Dunlin	Calidris alpina	0	0–10	0	1
Jack Snipe	Lymnocryptes minimus	0	0–2	0	1
Common Snipe	Gallinago gallinago	10–20	30–70	0	1
Eurasian Woodcock	Scolopax rusticola	0	0–3	0	1
Eurasian Curlew	Numenius arquata	0–1	0–20	1	1
Common Sandpiper	Actitis hypoleucos	0	0–5	0	1
Green Sandpiper	Tringa ochropus	15–20	40–100	0	1
Black-legged Kittiwake	Rissa tridactyla	0	2–15	0	1
Black-headed Gull	Larus ridibundus	10 000–17 000	80 000–140 000	0	1
Little Gull	Larus minutus	0–1	0–3	1	1
Common Gull	Larus canus	700–1900	600–1300	1	0
large gulls	Larus argentatus/cachinans	9–20	10–30	1	1
Great Black-backed Gull	Larus marinus	0	0–1	0	1
Western Marsh Harrier	Circus aeruginosus	0	0–5	0	1
Hen Harrier	Circus cyaneus	0–3	700–1500	0	1
Montagu's Harrier	Circus pygargus	0	0–1	0	1
White-tailed Eagle	Haliaeetus albicilla	10–15	40–80	0	0
Osprey	Pandion haliaetus	0	0–3	0	1
Common Kingfisher	Alcedo atthis	75–110	400–800	0	1
Meadow Pipit	Anthus pratensis	0	1500–3000	0	1
Grey Wagtail	Motacilla cinerea	7–8	500–1000	0	1
White Wagtail	Motacilla alba	0	700–1400	0	1
White-throated Dipper	Cinclus cinclus	290–340	1800–3500	0	0
Bearded Tit	Panurus biarmicus	0	0–60	0	1
Eurasian Penduline Tit	Remiz pendulinus	0	0–4	0	1
Reed Bunting	Emberiza schoeniclus	0–7	800–1600	0	1

Late-departing/early-arriving migrants coded as 1 and others (resident species) coded as 0.

The results for all of the alternative model specifications (Table 2) are consistent. No matter whether the overlap (Fig. 2) indicator or overlap proportions were used, and regardless of the source of species' distribution and population size data (IWC or AWB), the only variable with a significant effect was migratory strategy. Species which can be considered as late-departing (December) and/or early-arriving (February) migrants (see Methods and Table 1) tended to exhibit, on average, a markedly lower presence of overlap. Furthermore, migratory strategy was the only variable with coefficient estimates that did not change sign throughout all considered models. An overlap between IWC and AWB estimates was recorded in 31% (13 of 42 species) of late-departing/early-arriving migrants and 82% (9 of 11 species) of resident species. Among the resident species, IWC numbers were lower in comparison to AWB data in two widespread species, the White-tailed Eagle H. albicilla and White-throated Dipper C. cinclus. In contrast, population size estimated using IWC data was higher than the AWB population estimate in Ferruginous Duck (Table 1). Wintering numbers of this species were estimated to be between 7 and 20 individuals and their occurrence is limited to relatively few sites mostly well covered by the IWC.

Table 2. GLM analysis of effect of migratory strategy and of population distribution and population size on overlap between IWC and AWB population estimates for n = 81 species in all models. Distribution and population size obtained from IWC data in (a) and AWB data in (b); parameter estimates are presented with robust standard errors in parentheses; * P < 0.05, ** P < 0.01, *** P < 0.001.

	Overlap		Overlap proportion (AWB)		Overlap proportion (IWC)
	(a)	(b)	(a)	(b)	(a)	(b)
Migratory strategy	−3.07*** (0.86)	−3.45*** (0.80)	−1.87*** (0.50)	−2.15*** (0.53)	−2.20*** (0.67)	−2.63*** (0.63)
Distribution	−0.40 (1.20)	−1.16 (0.97)	−2.25 (1.20)	−1.02 (0.83)	0.30 (1.00)	−1.59 (0.92)
Population size	0.45 (0.60)	0.58 (0.59)	1.12 (0.60)	0.34 (0.47)	−0.10 (0.59)	0.75 (0.56)
Constant	1.20 (0.84)	2.18** (0.85)	−1.24*** (0.36)	−0.28 (0.66)	0.34 (0.68)	1.39 (0.76)

Finally, actual wintering population sizes were estimated for all 79 species recently recorded by IWC between 2009 and 2013 (Table 3).

Table 3. Population size estimates according to data from the IWC between 2009 and 2013.

Common name	Species	Population size	1% Threshold
Mute Swan	Cygnus olor	3000–4000	35
Whooper Swan	Cygnus cygnus	0–15	0
Swan Goose	Anser cygnoides	0–1	0
Bean Goose	Anser fabalis	1100–14 000	70
Pink-footed Goose	Anser brachyrhynchos	0–3	0
White-fronted Goose	Anser albifrons	6000–58 000	240
Lesser White-fronted Goose	Anser erythropus	0–2	0
Greylag Goose	Anser anser	1900–5700	40
Bar-headed Goose	Anser indicus	0–1	0
Canada Goose	Branta canadensis	0–4	0
Barnacle Goose	Branta leucopsis	0–20	0
Red-breasted Goose	Branta ruficollis	0–6	0
Egyptian Goose	Alopochen aegyptiaca	0–8	0
Ruddy Shelduck	Tadorna ferruginea	0–3	0
Shelduck	Tadorna tadorna	0–1	0
Muscovy Duck	Cairina moschata	0–17	0
Ringed Teal	Callonetta leucophrys	0–2	0
Wood Duck	Aix sponsa	2–15	0
Mandarin Duck	Aix galericulata	0–36	0
Eurasian Wigeon	Anas penelope	100–160	1
American Wigeon	Anas americana	0–1	0
Chiloe Wigeon	Anas sibilatrix	0–1	0
Gadwall	Anas strepera	65–170	1
Common Teal	Anas crecca	460–690	6
Mallard	Anas platyrhynchos	162 000–194 000	1800
Northern Pintail	Anas acuta	10–15	0
Northern Shoveler	Anas clypeata	3–20	0
Red-crested Pochard	Netta rufina	0–6	0
Common Pochard	Aythya ferina	800–2000	15
Ferruginous Duck	Aythya nyroca	0–9	0
Tufted Duck	Aythya fuligula	4100–5800	50
Greater Scaup	Aythya marila	20–100	0
Common Eider	Sommateria mollissima	0–1	0
King Eider	Clangula hyemalis	0–6	0
Common Scoter	Melanitta nigra	0–3	0
Velvet Scoter	Melanitta fusca	15–115	0
Common Goldeneye	Bucephala clangula	850–1300	10
Great Smew	Mergellus albellus	70–120	0
Red-breasted Merganser	Mergus serrator	1–8	0
Goosander	Mergus merganser	2800–4000	35
Ruddy Duck	Oxyura jamaicensis	0–1	0
Red-throated Diver	Gavia stellata	0–4	0
Black-throated Diver	Gavia arctica	3–9	0
Great Northern Diver	Gavia immer	0–2	0
Little Grebe	Tachybaptus ruficollis	470–1000	7
Great Crested Grebe	Podiceps cristatus	140–260	2
Red-necked Grebe	Podiceps grisegena	0–1	0
Slavonian Grebe	Podiceps auritus	0–8	0
Black-necked Grebe	Podiceps nigricollis	0–2	0
Great Cormorant	Phalacrocorax carbo	10 600–14 100	120
Pygmy Cormorant	Phalacrocorax pygmeus	0–1	0
Great Bittern	Botaurus stellaris	1–7	0
Little Egret	Egretta garzetta	0–1	0
Great White Egret	Egretta alba	260–630	4
Grey Heron	Ardea cinerea	2000–3000	25
White Stork	Ciconia ciconia	0–3	0
Water Rail	Rallus aquaticus	1–5	0
Common Moorhen	Gallinula chloropus	390–800	5
Common Coot	Fulica atra	9200–12 800	110
Eurasian Crane	Grus grus	0–1	0
Northern Lapwing	Vanellus vanellus	9–20	0
Jack Snipe	Lymnocryptes minimus	0–1	0
Common Snipe	Gallinago gallinago	0–4	0
Green Sandpiper	Tringa ochropus	9–25	0
Eurasian Curlew	Numenius arquata	0–1	0
Black-headed Gull	Larus ridibundus	8300–10 500	95
Mediterranean Gull	Larus melanocephalus	0–1	0
Little Gull	Larus minutus	0–3	0
Common Gull	Larus canus	990–2900	25
Lesser Black-backed Gull	Larus fuscus	0–1	0
large gulls	Larus argentatus/cachinans	470–2500	10
Marsh Harrier	Circus aeruginosus	0–1	0
White-tailed Eagle	Haliaeetus albicilla	120–200	2
Osprey	Pandion haliaetus	0–1	0
Common Kingfisher	Alcedo atthis	80–250	2
Grey Wagtail	Motacilla cinerea	25–50	0
Pied Wagtail	Motacilla alba	15–35	0
White-throated Dipper	Cinclus cinclus	400–730	6
Bearded Tit	Panurus biarmicus	0–22	0

DISCUSSION

In this study, long-term data from the IWC (1966–2013) were used to estimate wintering population size between 1982 and 1985 (i.e. the period covered by AWB), including imputed data from missing site counts. This latter procedure reduces the disadvantage of using IWC monitoring with respect to missing data (Fouque et al. 2009, Musil et al. 2011, Lehikoinen et al. 2013), considered as one of the factors affecting the accuracy of estimates (Kershaw & Cranswick 2003b) and is commonly used by many European bird monitoring schemes (Gregory et al. 2007, Lehikoinen et al. 2013). After minimizing this disadvantage of IWC data, we were able to consider the differences between wetland-site related IWC estimates and grid square AWB estimates without one possible source of census errors (i.e. the difference from the true value of the estimate – Bibby et al. 1992). The concept of precision is very important, especially in designing a bird monitoring study (Bibby et al. 1992) and the long-term IWC series used helped to produce more accurate IWC estimates of wintering population sizes in the AWB period.

Our study showed that the differences in IWC and AWB were not significantly affected by population size and species distribution. Therefore, the prediction of density-dependent bias of the estimates was not confirmed. The methodology of the AWB with its complete coverage of grid squares involving also non-wetland areas was not generally found to be more suitable for wintering waterbird estimates either for more distributed or rare species and did not provide any useful supplementary information for use in the calculation of national population estimates. This finding is in contrast with the study of Jackson et al. (2006) which found differences between Wetland Bird Surveys and Dispersed Waterbird Surveys. Nevertheless, when we take in to account individual species, the approach of AWB could improve the national population estimates in White-tailed Eagle and White-throated Dipper, species which predominantly occur in low numbers at many sites. By contrast, numbers of White-tailed Eagles and White-throated Dippers have been increasing in recent decades in the Czech Republic and their range has expanded – an effect more pronounced in White-throated Dipper (Musil et al. 2011). We can therefore expect even lower relative numbers of both species using IWC data in the future, assuming annual counts at the same sites, and it may be necessary to extrapolate numbers derived from IWC counts to produce more accurate estimates for these two species. Similarly, the population size of White-throated Dippers has been increasing with the recent milder winters in Norway (Nilsson et al. 2011) and the effect of mild winters could also cause changes in the winter distribution of White-throated Dippers in Czech Republic with an expansion of range to unfrozen sites.

Migration strategy had a significant effect when comparing IWC and AWB national population size estimates. In the case of the Czech Republic, counts carried out in December and February also cover late-departing (December) or early-arriving migrants (February) of some waterbird species, i.e. species whose breeding population size and/or numbers occurring during migration and/or breeding season exceed their wintering numbers (Hudec 1994, Hudec et al. 1999, Musil et al. 2001). These species can be recorded in December (late-departing migrants) and/or in February (early-arriving migrants) in numbers that exceed wintering population sizes. Therefore, AWB estimates exceed IWC estimates in late-departing or early-arriving migrants.

In general, the estimates of wintering populations of waterbirds in Europe are based on mid-January counts when numbers are most stable, with an emphasis of annually repeated monitoring of the same sites (Delany et al. 1999, Wetlands International 2006, 2013). There is a long tradition of monitoring internationally and nationally important sites which no doubt serves as a relevant tool for waterbird population estimates (Gilissen et al. 2002, Wetlands International 2006). Nevertheless, it is worth considering that the January national population estimates of late-departing or early-arriving migrants should not reach the peak numbers of individuals present in the Czech Republic during the course of a winter season. Correspondingly, Kershaw & Cranswick (2003b) presented the peak numbers in winter for population size estimates (Underhill & Prŷs-Jones 1994, Atkinson et al. 2006).

Encouragingly neither the pattern in species distribution nor the population sizes were responsible for differences between IWC and AWB estimates. Although based on a sample from one individual country's data, we support the relevance of the IWC methodology to produce reliable estimates of wintering waterbird numbers, probably because local conditions in mid-January are likely to constrain wintering waterbirds species to concentrate in a limited number of sites. Reliable national estimates constitute an essential tool for calculating the 1% thresholds for waterbird species and can indicate nationally important wetland sites for future conservation and protection. Knowledge of population size is fundamental for the application of conservation planning and action (Sinclair et al. 2006, Jackson et al. 2009) at both the national and international levels. Nevertheless, studies that compare the different methodologies used in estimating wintering waterbird numbers are either scarce or based on a relatively small geographical scale, usually at the regional or country level (Kershaw & Cranswick 2003a, Jackson et al. 2006). Here we are able to document, by comparing different methodologies, that the IWC can produce reliable national estimates of most species and so provide a comprehensive basis for management and decision-making, contributing towards the conservation of waterbirds on a flyway level (Boere & Stroud 2006, Hagemeijer 2006).

ACKNOWLEDGEMENTS

We are very grateful to all those volunteers who have been involved in both IWC and AWB counts as well as to the census co-ordinators of the International Waterbird Census (IWC) in the Czech Republic (i.e. Bohuslav Urbánek, Vladimír Fiala, Čestmír Folk, Josef Křen, Ivana Kožená and Jitka Pellantová). Moreover, our thanks must go to the Czech Society for Ornithology for help with organizing IWC in the Czech Republic. We are also grateful to Steve Ridgill for language improvements.

FUNDING

This work was supported by the Ministry of Education, Youth and Sport, Project C 22: ‘Support of PhD students and post-docs in University of Economics and Czech University of Life Sciences in Prague’.