METHODS

 The methods were similar to those of the 1985 survey (Williams 1986), with observa-tion (using binoculars and telescopes) providing the means of data collection. The great majority of passing migrants noted during the surveys were recorded during pe-riods of prolonged observation from suitable vantage points, notably the Lotus Hills. Daily counts were made of migrants present at areas representing the various habitats in and around Beidaihe.
  The 1986 and Earthwatch surveys placed special emphasis on recording actively mi-grating birds, through maintaining observations from vantage points at the Lotus Hills, at the western edge of town. This was because the totals of passing migrants re-corded at Beidaihe can be significant, and comparison of results from different seasons is relatively straightforward.
  During the other survey periods, there was also good coverage from the Lotus Hills watchpoints, though this was less systematic.
  During the 1986 and Earthwatch surveys, records of birds at the various localities in and around Beidaihe were logged. Otherwise, to-tals of birds present in or migrating over the Beidaihe area were recorded.

Methods used to record passing migrants

The main locality for recording passing mi-grants was the Lotus Hills. In spring 1985, observations were made from the southeast-ern hill. This is a little lower than the ‘main’ hill, to the northwest, and views to the west and northwest are obscured. However, there are excellent views in other directions. Soon after members of the 1986 team began obser-vations began here in autumn 1986, they were waved away by soldiers (the watchpoint is near a military area). Other sites on the main hill were tried, but none proved satisfactory— trees obscured views and there were too many visitors, and a rocky outcrop west of the main hill was chosen as the watchpoint for much of the autumn. This is lower than the southeast-ern hill, and views to the east are poor as they are blocked by the main hill (most of Beidaihe cannot be seen), but views over the plain are excellent.
  This outcrop proved a good vantage point, but in 1986 most cranes passed to the east of the Lotus Hills, with flocks disappearing be-hind the main hill for long periods, or only located as they appeared over the southern slope of the hill, often too distant for identifi-cation. Largely to obtain better views of the cranes (and possibly record more, and iden-tify a higher proportion), observations were made from the southeastern hill in late autumn 1987, and during the Earthwatch sur-veys (there was no trouble from soldiers). The rocky outcrop was regularly used as a watchpoint in 1987 until 20 October and in 1988 to around mid-October, and occasion-ally thereafter; Hornskov (1989) considers it better than the southeast hill for recording migrating small passerines. It is probably also better for recording the birds which mainly pass over the plain, e.g. raptors and storks.
  Greater care was taken with counts and es-timates of numbers of birds in flocks of cranes and storks than, say, Rooks Corvus frugilegus or Carrion Crows C. corone. Thus flocks of Rooks or Carrion Crows were typi-cally ‘counted’ by one observer, who might make a fairly quick estimate of, say, 60 birds (perhaps by counting ten birds, then estimat-ing how many tens the flock contained), which was entered in the log. Crane and stork flocks were often counted by more than one observer, the figures only being announced once all counts were complete. In some cases, perhaps after re-counts, the figures agreed exactly and were entered in the log. In others, a ‘best’ figure was chosen, or an average was entered. We found that counts might produce the same figure for even the larger flocks of cranes, perhaps with over 300 birds (the larg-est ‘flocks’ are usually groups of crane formations, each of which can be counted separately). This suggests that even the larg-est crane flocks can be counted accurately. Problems can arise, however, on ‘wave’ days, when flocks may pass in quick succession, not allowing time to repeat counts.
  Whenever possible, flocks of birds which use thermals to gain height and so minimise energy use during migration were not counted as they ‘thermalled’, as at this time they are probably impossible to count with any accu-racy (cranes, storks, raptors, swifts and crows all use thermals in this manner). If the birds were in a thermal when they were located, we found it best to wait until they began heading off from the top of the thermal (as raptors do), or re-grouped into the flock formation (e.g. cranes) before we began counting.
  To minimise over-recording of passing large birds such as storks, cranes and raptors by entering records in the log for the same birds seen from more than one locality, numbers of birds, flight paths and times were compared. If it appeared that birds had been seen from more than one locality, only one observation was entered in the log. For small birds, espe-cially passerines, all observations were logged as it was assumed that the great majority of those seen from the Sandflats would be dif-ferent to those seen from the Lotus Hills.
  During the 1986 and Earthwatch surveys, totals entered in the log were simply the sums of all counts. In 1987 and 1988 (Hornskov 1989) the totals were rounded up or down somewhat arbitrarily if it was felt this better reflected the true accuracy of the figure, which might be largely derived from approximate counts, e.g. a total of 232 might be rounded down to 230.

Variations in the degree of daily coverage

  The following factors influenced the degree of daily coverage.

Number of observers  The number of ob-servers during the surveys fluctuated, ranging from one to ten or more (typically three to seven). The main periods with one observer were 8 September to 7 October 1988, 14-29 September 1989 and 19 August to 22 October 1990 (though, especially in 1990, there was some help from birdwatchers who visited during these periods), and the last ten days of November 1987.

Variation in weather and numbers of birds grounded or passing  As in spring 1985, the weather was rarely solely responsi-ble for affecting the degree of coverage of the area: there were a few days when heavy rain or snow prompted observers to shelter in ho-tel rooms (considerable migration was sometimes evident once the weather began clearing—i.e. a cold front was moving away to the east).
  In 1986, the Lotus Hills observations were made daily, weather permitting. The watch-point was manned on a rota basis, with rarely less than two observers present; observation periods totalled 853.6 hours (2624 man-hours)—an average of around 9.25 hours per day. Initially, observations typically began by 07h30; the starting time became earlier as Pied Harriers Circus melanoleucos could be passing in numbers soon after dawn, and sev-eral watches in September began at 06h30 (Beijing summer time; just before sunrise). The observations typically ended by 15h00-16h00 (rarely 12h00-13h00 on very quiet days), once continuing to 18h40 (on 12 September, a day of heavy migration). Later in the autumn, observations tended to start and finish later (often 06h30-07h00 to 16h00-17h00, winter time: China has since abandoned summer time). This was because fewer birds were passing in the early morn-ing, and cranes, especially, were mainly seen in the afternoon. Again, observations were cut short if there was very little passing, pro-longed if there was heavy migration, and began earlier than normal if the weather seemed promising.
  Lotus Hills observations during the Earth-watch surveys roughly followed the late autumn pattern, though were more likely to be abandoned if little was passing and conditions were not promising, and there were no obser-vations at Beidaihe on a few days when other localities were visited (on all these days, the weather seemed poor for migration).