Disturbance and flock size changes in Greenland Whitefronted Geese wintering in Ireland

Transcription

1 Disturbance and flock size changes in Greenland Whitefronted Geese wintering in Ireland D.W. N O R R IS S and H.J. W IL SO N Introduction T h e G re e n la n d W h ite -fro n te d G o o se, A n se r albifrons flavirostris, has a northern and w estern w inter distribution in Ireland a n d B ritain, coincident w ith the original d istrib u tio n o f b o g lan d w h ere it tra d itionally fed. P o pulation estim ates fell from 17,500-23,000 in th e 1950s to 14,300-16,600 by 1979, b u t the decline had not been uniform th ro u g h o u t th e w in ter range. N um b ers in S cotland increased slightly and in W exford rem ain ed constant b u t the population declined by ab o u t 60% in th e w est and m idlands o f Ireland (R uttledge and O gilvie 1979). P rotection w as introduced in S cotland and th e R epublic o f Ireland in 1982, w hile voluntary bans operated in W ales and N o rth ern Irelan d from before this d ate. C onsequently the population had in creased to nearly 20,000 by spring M ost Irish flocks outside W exford still feed in sem i-natural habitats. R uttledge and O gilvie ( 1979) dscribed how flocks have b een p artially o r w holly constrained from leaving bogland by the sm all size and high d istu rb an ce levels on alternative grassland areas. H a b ita t loss, shooting and disturban ce w ere cited as th e m ost im portant facto rs in th e decline o f this segm ent of the p o p u la tio n. E lse w h e re m o st w in terin g flocks have ad ap te d to farm land feeding on large estates o r on islands and have benefitted from lim ited access, controlled shooting and large feeding areas w ithin w hich they could m ove w hen disturbed (R u ttled g e and O gilvie 1979). G eese are inefficient herbivores and m ust sp en d m ost w inter daylight hours feeding. C o n seq u en tly, disturbance w as th e m ost im p o rtan t facto r lim iting goose distribution on farm land (K uyken 1969; O w en 1972) and has b een given as the cause o f declines and d esertio n s of E u ro p ean W hite-fronted G eese, A.a. a lbifrons, in H ungary (S terbetz 1967) and B ritain (O gilvie 1968), w hilst access to large areas rarely visited by m an have b een described as decisive in d e te r m ining selection o f feeding areas by B ean G eese, A. fa b a lis, in so u th ern Sw eden (M athiasson 1963). A lthough drainage was th e com m onest reason recorded for desertion o f w et grassland sites by G reenland W h ite -fro n te d G e e se (R u ttle d g e and O gilvie 1979), O w en (1972) suggests th at increased agricultural disturbance resulting from d rainage is th e im p o rtan t factor rath er th an d rainage itself. O n th e positive side, ab o u t ten statu to ry no-shooting areas w ere d e c la re d fo r G re e n la n d W h ite -fro n te d G eese, m ainly for th e larger flocks, during th e 1970s. D uring the w inters to , th e W ildlife Service organised a survey of th e num bers and distribution of G reenland W hite-fronted G eese as p art o f a larger conservation program m e in Ireland and B ritain. T he status and w inter distribution have already been described in a series of annual reports by the G reenland W hitefronted G oose Survey, A berystw yth and the W ildlife Service, D ublin. In this paper disturbance data from standardised survey cards is analysed to see w hether changes in flock size w ere influenced by disturbance. Methods Field visits w ere m ade at least m onthly during the w inters and F o r th e d u ratio n o f a visit o b serv ers recorded the length of tim e geese w ere present, the num ber and type of disturbances and w hether disturbed geese left the site. T he area o f a feeding site is delim ited by all the recorded observations o r field signs of geese w ithin 1 km of each other. Feeding in larger areas tend to reduce flying tim e and energy costs since disturbed geese can m ove short distances w ithin a site, w hile having a large num ber of sites from w hich to choose affords m ore com prehensive protection from locally adverse feeding conditions. T he energetic costs o f disturbance are a function of both the disturbance rate and the degree to w hich the num ber and size of the feeding sites within a flock s range can protect geese from disturbance. These elements can be quantified by a disturbance index (Ps/R ) for each site, w here Ps is th e probability of geese staying at a feeding site after a disturbance and R is the disturbance rate. A disturbance was 63 W ildfowl 39 (19NS): 63-70

2 64 D.W. N orriss and H.J. Wilson included in the analysis only w hen it caused geese to take flight and w here th ere was an observed cause of the incident. D isturbances caused by observers w ere excluded, because visits w ere norm ally lim ited to one p er m onth and th e calculation of disturbance rates is unlikely to have been significantly affected. The index was then sum m ed for each site for w hich an arbitrary figure of m ore th an five disturbances w ere observed and at w hich geese w ere present for m ore than 5% of the total tim e observed at all sites w ithin a flock s range. T he sum m ed total gave a disturbance index for each flock, although insufficient data w ere available for all flocks. Favourable ranges w here feeding sites are large and geese are well shielded from disturbance are represented by larger index values, unfavourable ranges by sm aller ones. In o rd er to relate num erical changes of flocks to th e quality of the feeding range in th e period 1950s to (w hen no com parable data on disturbance had been collected), flocks w ere categorised prim arily according to th e num ber and size of feeding sites w ithin th eir feeding range. These categories are defined as follows:- A. R anges w ith m ore than 10 feeding sites, one o r m ore being over 400 ha. O n various grassland habitats. Partial p ro tection o f all flocks by no-shooting areas o r private estates. B. R anges w ith 3 or m ore feeding sites, each site usually u nder ha, m axim um 250 ha. O n w et grassland o r callows. Largely unprotected. C. R anges w ith 1 or 2 feeding sites, each site u nder ha, on a variety of habitats. L argely unprotected. Results Change in flo ck size from to Figure 1 relates the changes in m axim um flock size to disturbance indices in 18 flocks in the west and m idlands of Ireland betw een and M axim um counts for both w inters are recorded in th e A ppendix. C hange in flock size was significantly correlated to the disturbance index (rs=0.5(), d f= 16, P < 0.0 5, S pearm an s rank correlation). H ow ever, the relationship was considerably affected by one flock w ith an D isturbance index Best 12 Figure 1. Relationship of change in flock size, measured by % change of maximum count between and , to a disturbance index during the same period. Each point represents one flock; data restricted to Irish flocks outside W exford. Flocks primarily on dry pastures shown by, on wet pastures by + and on bogland by A. For m ethod of calculating disturbance index and for statistics see text.

3 D isturbance o f geese in Ireland 65 unusually sm all disturbance index value (A in Figure 1). Several feeding sites of flock A have becom e know n since th e period on w hich this analysis was based and one or m ore m ajor feeding sites probably still rem ain undiscovered to judge by the frequency w ith which this flock has been m issed during counts. T hus its disturbance index was underestim ated. W hen this flock is excluded from the analysis, a significantly im proved correlation betw een change in flock size and disturbance index results (rs= 0.63, d f= 15, PcO.O l). T he disturbance index value of the in tercept of the regression line on the x-axis for a particular change in flock status provides a q u a n tifie d m a n a g e m e n t o b je c tiv e. H ow ever, th ere are two unrelated problem s with fitting a regression line. T he first is the non-norm al distribution o f both sets of data. T he points represent m eans of three years data and the points them selves can be assum ed to be norm ally distributed about the x and y axes. Secondly, the correlation is not described by a straight line, but the num ber o f points are too few to define their distribution and to m ake the apropriate transform ation. F o r practical purp o ses, th e regression eq u atio n was repeatedly calculated as flocks o f larger and larger disturbance index value w ere added one by one. T he estim ate of th e required intercept th at is m ost accurate and econom ic is given by the regression line with the steepest slope and narrow est confidence limits. This was percentage change in flock status = x (± 2.4 0, 95% confidence lim its), n = 8. D isturbance rates T able 1 shows th at overall disturbance levels are m uch higher on dry grasslands, w hether sem i-natural o r reseeds, than on w et grasslands, callows and bogland. Farm ing activities are prim arily responsible for th e th reefold difference in disturbance levels. H eavier disturbance on drier soils results from a larger proportion of arable land and m ore freq u en tly outw in tered stock, w hereas w et grasslands in w inter are generally too fragile for stock o r m achinery. A t first th e higher overall disturbance levels on dry grassland seem at odds w ith th e general tren d for G reen lan d W hitefro n ted G eese to m ove to m ore intensely farm ed areas. B ut th ere field and farm sizes a re larger and d istu rb ed geese are m ore likely to m ove w ithin sites. A lso goose flocks are often eith er p ro tected (even th o u g h th e prim ary m otive m ay be stock p ro tectio n ), o r shooting is m anaged and d istu rb an ce is lim ited by restricting public access. P ro tected geese respond less freq u en tly and less intensely to disturbances and th ey h abitu ate fairly quickly to farm ing Table 1. Variation, by habitat, in the type, importance (top row, % occurrence) and rate (bottom row, no./hr) of disturbance during and after the shooting season. D ata for individual sites in flock ranges in the west and midlands of Ireland to Disturbance due to: Arrival to 31 January Farming Shooting Aircraft Other n Total 1 February to Departure Farming Shooting Aircraft Other n Total Dry 56% 10% 15% 19% 4! 49% 3% 12% 36% 33 grasslands h h Wet grass 48% 22% 10% 20% 40 32% 16% 26% 26% 19 land/callows Blanket 22% 11% 0% 67% 9 50% 0% 25% 25% 4 bog a Causes of disturbance in other category, by habitat; dry grasslands - 2 raptor, 2 fox, 1 dog, 1 thunder, 4 horse riding, 1 fox hunting, 2 pedestrian, 5 boat, 1 gamekeeping, 1 fisherman (6 nonhum an. 14 hum an); wet grassland/callows - 1 dog, 1 raptor, 1 deer, 3 boat, 3 traffic, 2 fishermen, 2 gamekeeping (4 non-human, 10 hum an); blanket bog - 1 fox. 1 deer, 1 forestry, 1 turf-cutting, 1 boat, 1 traffic, 1 pedestrian (2 non-human, 5 human). b As there was no significant difference in disturbance rates during and after the shooting season (ts=1.131, df=16, ns), the mean value for the winter period has been used in the calculation of rates for each type of disturbance.

4 66 D. W. N orriss an d H.J. Wilson disturbance. T hus a shift of feeding areas to extensive farm land can give a larger value o f the disturbance index even w ithout reduction of agricultural disturbance. In fact significantly less disturbance was recorded from estates and naturally protected islands than from o th er farm s in the dry grassland category (T able 2). This is largely because m ilking herds are usually overw intered in sheds and such agricultural activity as there is on the fields is often m echanised. The shooting com ponent of disturbance rates is o f sim ilar absolute value in dry and w et grasslands b u t is o f g reater relative im p o rtan ce in th e latter (T able 1). O n average sh ooting caused a q u a rte r o f all w et grassland distu rb an ces, b u t in wildfowling areas it could be m uch m ore im portant. H eavy sh ooting pressure, com bined with sm all size o f w et grassland feeding sites, w as su sp ected as causing the n ear desertion by fo u r flocks in C o. C lare by and fo r th e ir su b seq u en t rapid recolonisation a fte r th e shooting m oratorium was introduced. Table 2. The probability of disturbed geese staying at a feeding site (Ps) and the disturbance rate (R), (means with 95% confidence limits), in different habitats. D ata for individual sites in Rest of Ireland' flock ranges, to Habitat Ps R (no./hr) Dry grasslands 0.396±0.154 n = ±0.235 n = 14 Wet grasslands and callows 0.269±0.124 n= ±0.077 n=20 Islands and areas of limited access ±0.064 n = 7 Blanket bog 0.048±0.047 n= ±0.047 n=13 C hanges in status fr o m the 1950s to W h en flocks are grouped by the num ber and size o f feeding sites, a sim ilar p attern o f flock declines and extinctions in relation to quality o f th e feeding range is apparent betw een the 1950s and as was fo u n d in th e p resen t study (Figure 2 and A p p en d ix ). F locks w ith th e best feeding ranges (category A ) have m oved to extensive areas o f farm land o r callows. W hile som e can be difficult to co u n t accurately, their status a p p ears to have changed relatively little since th e 1950s. R uttledge and O g'lvie (1979) w ere concerned at the loss of raised bogs w hich w ere used for feeding or as refuges. H o w ev er, the concurrent creation extinct extinct extinct / decline ^decline / deci Ine Figure 2. Change of flock status from 1950s to in relation to feeding range characteristics, west and midlands of Ireland. Category A flocks (left), B (centre) and C (right). Status change was calculated as percentage change in maximum counts between the two periods; data for the 1950s from Ruttledge and Ogilvie (1979).

5 Disturbance o f geese in Ireland bl o f no-shooting areas for m any of the larger flocks has so far evidently com pensated for th e loss o f bogs since these flocks have increased since T h e flocks m ost th re a te n e d by drainage w ere th o se w ith a num ber o f sm all, w et grassland feeding sites (category B ). O verall these flocks have had the highest rate of decline. O f the seven Irish flocks which b ecam e extinct during this period (T able 3 ), five w ere o f this type an d th ree w ere affected by drainage (R u ttled g e and Ogilvie 1979). M any category B flocks used turlough areas in th e m id-w est and w ere also su b ject to high levels o f shooting disturb ance. P ro tectio n has reversed th eir fo rtu nes and these flocks are now stable or increasing. Table 3. The number of Greenland Whitefronted Goose flocks that have become extinct since the 1950s and the number currently extant. The num ber of extinct Irish flocks has been estim ated from data on the num ber and distribution of deserted haunts, compiled by Ruttledge and Ogilvie (1979). D ata for Scotland from G reenland W hite-fronted Goose Survey annual reports. Extinct Extant Period 1950s Scotland Ireland Flocks w ith the sm allest ranges (C ategory C) have also fared badly, particularly w here the num ber of alternative feeding sites is also small. This category includes the o th er tw o flock extinctions. T he original bogland range and the grassland areas to which these flocks m oved afforded little protection to in c re a se d d is tu rb a n c e levels. A rte ria l drainage and shooting have not been im portan t pressures but small scale agricultural developm ent and increased recreational disturbance have continued to erode their m arginal suitability. A t R ahasane, C o. G alw ay, for instance, deeper flooding of the feeding site, resulting from extensive field drainage upstream, periodically restricts the W hitefront s feeding area to the m ore disturbed periphery. A ll flocks in this category have recently show n net declines in num bers. Discussion D isturbance effects It is ap p aren t from Figure 1 that m uch of th e variation in flock size changes was associated with disturbance levels. D isturbance effects could operate in two ways. Firstly, a sm aller disturbance index may favour sm aller flocks if as a result their tolerance of disturbance increases. O w ens (1977) found that the distances at which disturbed B rent G eese, Branta bernicla, w ere put to flight decreased w ith sm aller flock size and suggested this could be one reason why flocks feeding in narrow creeks w ere small. M ore im portantly, disturbance levels directly influenced th e energetic costs of feeding, and hence the suitability o f a site, by increasing flying tim e and reducing tim e available for feeding (ef. W hite- R obinson 1982). D isturbance-m ediated declines m ay be caused by increased m ortality, reduced breeding success, em igration, or by any com bination of these. It m ight be expected th at significant levels of disturbance w ould affect breeding perform ance by reducing body condition, but th ere was no correlation betw een disturbance index and the proportion of juveniles in each flock, expressed as a percentage of the w inter m ean (S pearm an s rank correlation coefficient = 0.0, d f= 9, ns). H ow ever, the lack of an observed relationship does not m ean that disturbance levels have no effect on subsequent recruitm ent since successful b reed ers m ight redistribute them selves betw een w inters and such a p attern w ould then be m asked. T here is som e indication from age counts th at a redistribution o f fam ilies does in fact happen, locally at least, but data on m ovem ents of m arked birds are so far too lim ited to draw any conclusions. Vulnerability to disturbance C om parative data on disturbance rates of geese are scarce. G reenland W hite-fronted G eese on callows of a rem ote Shannon tributory m ade disturbance flights/ hour (calculated from data in M ayes, 1985), close to the m ean figure fo r sites of lim ited access recorded by this study. B rent G eese in SE England w ere much m ore disturbed, the m ean num ber o f disturbance flights/ hour varying betw een 0.28 in the quietest sites to 2.4 at w eekends in the m ost dis-

6 68 D. W. N orriss and H.J. Wilson turbed areas (O w ens 1977). In Essex, B rent G eese m ade significantly m ore frequent disturbance flights (although o f shorter d u ration) w hen feeding on saltm arshes c o m p a re d w ith fe e d in g on fa rm la n d (W hite-r obinson 1982). G reenland W hitefronted G eese w intering in Ireland generally had low disturbance levels, m ean rates varying from 0.06 to 0.51 flights/hour according to habitat (T able 2). Even so, the sm all disturbance indices of m any category B and C flocks will not allow geese to to lerate fu rth er deterioration in range quality (Figure 1). A substantial increase in agricultural disturbance levels following drainage, com parable to the difference betw een w et and dry grasslands recorded in T able 2 w ould cause the geese to desert, as suggested by O w en (1972). This is particularly so w hen several sites within a catchm ent are affected sim ultaneously by an arterial drainage schem e. T he vulnerability of these flocks is prim arily due to the small size o f individual feeding sites and is com pounded w hen few altern a tive feeding areas w ere available. This relationship betw een drainage, disturbance and food resource dispersion explains the higher extinction rates of Irish flocks generally and of C ategory B flocks in particular. D rainage has been an im portant factor in flock extinctions. T he desertion by three out of four category B flocks on w et grassland sites coincided w ith drainage (R u t tledge and Ogilvie 1979) but none of the six extant flocks ranges w ere arterially drained. Tw o points follow from this. U n p rotected flocks w ere unable to cope with the additional disturbance pressures resulting from arterial drainage, and all becam e extinct. Secondly, num bers have rem ained stable o r increased since protection in flocks w hose feeding grounds have not been arterially drained, while agricultural disturbance levels have rem ained unchanged. This im plies th at shooting pressures w ere the principal factor in earlier declines, although not apparently causing any flock extinctions. T olerance of disturbance m ay also vary according to diet. Figure 1 suggests that flocks on dry and im proved pastures have a g reater tolerance than those on w et pastures, but sam ple sizes are sm all. G eese that feed on high energy w aste from cereal and root crops show a g reater w ariness and to le ra n c e o f d istu rb a n c e th a n grazers (O w en 1972). Increased digestibility of grasses, characteristic of b etter grasslands, can be expected to confer the sam e advantage. M anagem ent prospects Flocks of G reenland W hite-fronted G eese on sem i-natural habitats have responded to protection and the creation o f no-shooting areas by heavier use of favoured feeding sites. A m ongst category C flocks for instance, m ean attendance at m ajor feeding sites (expressed as the % of visits th at geese w ere present) has shown a significant increase from 74% to 84% in the five years since protection (one-tailed t-test for paired sam ples, t= 2.8 3, d f= 5, P < ). Even w hen goose num bers involved are less than 1 0 0, problem s w ith farm ers have occurred w here geese are using im proved grassland. T here has been no proportional increase in use of reseeded grassland by W hitefronts in the last five years, but com plaints about dam age are likely to increase as num bers continue to grow. P ro tectio n from shooting and d isturb an ce have led to increases in m ost flocks on m ed iu m and good q u ality feeding ranges. Flocks on the R iver Shannon may be v u ln erab le to disturbance from shooting in w in ter a n d from pleasure cruisers and fishing in spring because o f the linear distrib u tio n o f feeding sites along the river; three o th e r flocks had ra th e r sm all disturbance indices (F igure 1). O therw ise present m easures ap p e a r to afford ad eq u ate protection from distu rb an ce pressures for these flocks. H ow ever, flocks in category C rem ain th re a te n e d. A ll seven flocks on ranges w ith one o r tw o sm all feeding sites have contin u ed to decline since These flocks are a conservation priority but it m ust be rem em b ered th eir sm all disturbance indices have b een calculated for the period a fte r th e shooting m oratorium started and im proved m an ag em en t beyond statutory p ro tectio n will be difficult to achieve. Even w ith successful m anagem ent, th eir future is d ep e n d a n t on the good will o f one o r two landow ners. Acknowledgements It is a pleasure to thank the team of 75 fieldworkers from the Wildlife Service (ex Forest and W ildlife Service), the Irish W ildbird Conservancy and the Irish Shoot Prom oter s Association in the Republic of Ireland and the D epartm ent

7 Disturbance o f geese in Ireland 69 of the E nvironm ent and Royal Society for the Protection of Birds, N orthern Ireland. C. M urphy, R. Nairn and N. Sharkey helped with adm inistration. O ur thanks to P. Dowding who kindly gave statistical advice, to P.J. W arner, A. Fox and D. Stroud for com m ents on an earlier draft and to T. O 'B rien who typed the manuscript. Summary An index of disturbance was calculated for wintering G reenland W hite-fronted Goose flocks based on observed disturbance rates and the refuge qualities of feeding ranges. In the three years following protection, changes in status of individual flocks were correlated with their disturbance indices. Agriculture was the single most im portant source of disturbance and overall rates of disturbance were highest on intensively m anaged land. However, the smallest indices (most disturbance) were calculated for flocks with few. small feeding sites. Such flocks are generally declining. Consequences of arterial drainage and shooting are discussed. Flocks with better quality feeding ranges are stable or increasing as a result of protection and are not currently threatened by arterial drainage. Prior to protection numerical trends of flocks showed a similar correlation with range quality. Differences in patterns of flock declines and extinctions before and after protection are consistent with known changes in disturbance pressures. Appendix Flock sizes used in the analyses. Num bers are maximum estimates for the 1950s (Ruttledge and Ogilvie 1979) and maximum spring counts for and (Wildlife Service records). Flock ranges are defined in Ruttledge and Ogilvie (1979). C A T E G O R Y A 1950s Ls. Foyle and Swilly L. Drum harlow L. Gara R. Suck Little Brosna Ls. Iron, Ennel C A T E G O R Y B Ls. Kilglass, Forbes L. Conn Rostaff, Altore Lr. L. Corrib Rahasane C arran, Lehinch Tullagher L. Derg R. Barrow (E) 50+ extinct (unknown causes) R. Suir (F) 200 extinct (arterial drainage) Killorglin (H) 50 extinct (field drainage) Clare R. (QT) 100+ extinct (arterial drainage) Loughglinn (UV) extinct (turbary) C A T E G O R Y C Bunduff L. Macnean Caledon Kilcolman N. Antrim Moors (B) unknown extinct (unknown causes) L. Key (Y) 50 extinct (afforestation) O TH ER Killarney Valley 69 52

8 70 D.W. N orriss and H.J. W ilson References K uyken, E Grazing of wild geese on grasslands at Dam m e, Belgium. W ildfowl 20: Lazarus, J Vigilance, flock size and domain of danger size in the W hite-fronted Goose. W ildfowl 29: M athiasson, S The Bean G oose, Anser fabalis (Latham ), in Skane, Sweden, with remarks on occurrence and migration through northern Europe. Acta Vertebratica 2 (3): Ogilvie, M.A The num bers and distribution of the European W hite-fronted Goose in Britain. Bird Study 15:2 15. O w en, M Some factors affecting food intake and selection in W hite-fronted Geese. J. Anim. Ecol. 41: Ow ens, N.W Responses of wintering Brent Geese to human disturbance. W ildfowl 28:5-14. R uttledge, R.F. and Ogilvie, M.A The past and current status of the G reenland Whitefronted G oose in Ireland and Britain. Irish Birds (3): S terbetz, I A M agyarorszagon telelölilikek ökológiái problém ai. Aquila 73 (4):33^t9. W hite-r obinson, R Inland and saltmarsh feeding of wintering Brent Geese in Essex. Wildfowl 33: D.W. Norriss and H.J. Wilson, Wildlife Service, Office of Public W orks, 2 Sidmonton Place, Bray, Co. W icklow, Ireland.