KERSTEN M, RAPPOLDT K & SMIT C (1981) On the accuracy of shorebird counts. LIMOSA 54 (2): 37-46.

This paper presents the results of some experiments carried out to estimate the accuracy of large-scale counts of shorebirds. The aim was to quantify both variation (consistency) and systematic errors (accuracy) in the figures yielded by these counts. Field experiments were conducted by 28 experienced observers on the Dutch Waddensea island of Ameland (figure 1). Additional information was obtained with slides showing groups of beans simulating bird flocks. The variation in the estimates of several observers for the number of birds in a certain flock is given as coefficient of variation (RSD).
      The average RSD for 17 flocks of roosting birds amounted to c. 20%. This figure holds for flocks of different species and is not affected by flock size (table 1). The RSD for flocks of birds flying towards a high tide roost does neither show a correlation with flock size (figure 2). The average RSD of the estimates of observers counting only Dunlins in these flying flocks amounted to 17%. The RSD of observers counting all species simultaneously exceeds 30%.
      Counts of beans on slides and of flocks of roosting birds are usually underestimates (figure 4, 5). The RSD of the number of birds counted by five observers in a certain area are given in figure 6. RSD's of species of which more than 100 birds were present amounted to c. 25% which is a little higher than for groups (table 1). For species occurring in small numbers the RSD's are larger. It is suggested that the chance that these birds are overlooked by some observers has a large effect on the size of the RSD. The results of shorebird counts on two successive days are compared in figure 7. This comparison was made assuming that no migration took place between these two counts. In many occasions RSD's of numerous species appeared to decrease to values below (!) 20%, which is the average value for the RSD of a single flock. The probable explanation for this difference is that random fluctuations in counting results neutralize each other when results of several bird flocks are added.
      In the Dutch Waddensea shorebirds are sometimes counted when they are sitting at the high tide roosts, sometimes when they are flying from the tidal flats to the roosting places. The results obtained by both methods counting the same birds are compared in table 4 showing no great systematic differences between the two counting methods. In areas where large numbers of different bird species occur the estimates by individual observers counting all these birds alone, are usually much lower then the estimates obtained by observers which had to count only the numbers of one or two species in the same area at the same time (figure 9). Eurasian Oystercatcher Haematopus ostralegus Dunlin Calidris alpina = Calidris alpina schinzii Eurasian Curlew Numenius arquata Common Redshank Tringa totanus European Golden Plover Pluvialis apricaria Red Knot Calidris canutus Grey Plover Pluvialis squatarola Ruddy Turnstone Arenaria interpres

[free pdf] [dutch summary]