Impacts of a volcanic eruption on the forest bird community of Montserrat, Lesser Antilles

Ibis

(2007)

149

298ndash312

copy 2007 The Authors Journal compilation copy 2007 British Ornithologistsrsquo Union

Blackwell Publishing Ltd

Impacts of a volcanic eruption on the forest bird community of Montserrat Lesser Antilles

B DALSGAARD

1

G M HILTON

2

G A L GRAY

3

L AYMER

3

J BOATSWAIN

3

J DALEY

3

C FENTON

3

J MARTIN

3

L MARTIN

3

P MURRAIN

4

W J ARENDT

5

D W GIBBONS

2

amp J M OLESEN

1

1

Department of Ecology and Genetics University of Aarhus Ny Munkegade Building 1540 DK-8000 Aarhus C Denmark

2

Royal Society for the Protection of Birds The Lodge Sandy Beds SG19 2DL UK

3

Forest amp Environment Division MALHE PO Box 272 Brades Montserrat

4

Montserrat National Trust PO Box 393 Olveston Montserrat

5

USDA Forest Service International Institute of Tropical Forestry Sabana Research Field Station

HC 2 Box 6205 Luquillo PR 00773

Volcanic eruptions are an important and natural source of catastrophic disturbance toecological communities However opportunities to study them are relatively rare Here wereport on the effects of the eruption of the Soufriegravere Hills volcano on the forest birdcommunity of the Lesser Antillean island of Montserrat The islandrsquos species-poor avifaunaincludes 11 restricted-range species including the Critically Endangered endemic MontserratOriole

Icterus oberi

Analysis of monitoring data from 1997 to 2005 indicates that countsof most species were substantially lower following major ashfalls However this effect wasshort-lived with rapid population recovery in subsequent years Furthermore levels ofseasonal rainfall appear to have been at least as important in determining population trendsas ashfall Overall most species were at least as abundant at the end of the study as at thestart and no forest bird species have been extirpated from Montserrat We discuss potentialecological drivers of ashfall impacts on populations there is some evidence that terrestrialforagers were most severely affected

Environmental catastrophes are predicted to have amajor influence on the survival of small isolatedanimal populations However catastrophic eventsare rare ephemeral and unpredictable and aretherefore difficult to study The bird communities ofthe Lesser Antilles Endemic Bird Area (EBA) areheavily shaped by such events The land area ofthe EBA is relatively small (6400 km

2

) comprisingnumerous small islands (Stattersfield

et al

1998)The current biogeography of Lesser Antillean birdshas emerged from this distribution of land massesbut has also been profoundly shaped by frequenthurricanes (Wauer amp Wunderle 1992 Wunderle

et al

1992 Wunderle 1995) and volcanic eruptions Mostislands in the Lesser Antilles are volcanic in originand during historical times major eruptions have been

moderately frequent with 16 volcanoes recorded asactive in the Holocene (wwwvolcanosiedugvp)A mass extinction related to explosive volcanism inthe Caribbean region occurred in the late Palaeocene(Bralower

et al

1997) The region is also one of thehottest of the worldrsquos biodiversity hotspots (Myers

et al

2000) and the EBA holds 24 restricted-rangebird species within the island chain there are 12single-island endemic bird species

Pyroclastic flows and explosive blasts of hot gasesand solids are characteristic of eruptions of Carib-bean andesite volcanoes and can cause completedestruction of biota in impacted areas However afar wider area can be affected by the ashfalls and acidrain that are also associated with such eruptionsThese do not cause total destruction of biota butmay still have a major effect on ecosystems Little isknown about the effects of volcanic eruptions onbird populations and almost all the information

Corresponding author Email GeoffHiltonrspborguk

copy 2007 The AuthorsJournal compilation copy 2007 British Ornithologistsrsquo Union

Volcanic impacts on forest birds

299

available comes from the eruption of Mount StHelens USA on 18 May 1980 (Hayward

et al

1982Andersen amp MacMahon 1986 Manuwal

et al

1987)Such information is relevant to the conservation of 15globally threatened bird species that are threatenedby volcanic activity and in particular the two birdspecies ndash Black-breasted Puffleg

Eriocnemis nigrivestris

and Mount Karthala White-eye

Zosterops mouroniensis

ndash that have their entire ranges on the slopes of currentlyactive volcanoes (BirdLife International 2004)

Here we examine the influence of the eruptionof the Soufriegravere Hills volcano on the forest birdcommunity in Montserrat (16

deg

N 62

deg

W 109 km

2

) aUK Overseas Territory at the northern end of theLesser Antilles One would predict that on smallislands volcanic eruptions might result in temporarypopulation bottlenecks and possibly extirpationwith recolonization occurring once the volcano hasfallen dormant The current eruption of the SoufriegravereHills volcano began in 1995 and continues at thetime of writing The volcano also erupted in 1640though information about this eruption is scantBriefly the eruption has been characterized by thegrowth of an andesite lava dome with associatedpyroclastic flows vulcanian explosion and debrisflows (wwwmvoms) Pyroclastic flows caused thealmost complete destruction of tropical forest inthe southern hill ranges of the island by late 1997In the remaining forested hill range the Centre Hillsashfalls and acid rains have been frequent andheavy during 1996ndash2005 interspersed with periodsof recovery This pattern of localized obliterationof ecological communities accompanied by muchmore widespread impacts of ashacid rainfall haspresumably been the pattern in previous eruptionsalthough direct evidence is lacking Montserratrsquosavifauna is of considerable conservation importanceas it includes the critically endangered single-islandendemic Montserrat Oriole (Hilton

et al

2003)the vulnerable four-island endemic Forest Thrushand nine other restricted-range species (Stattersfield

et al

1998 Arendt

et al

1999 Hilton

et al

2001)(see Table 1 for scientific names of the study species)

Regular forest bird monitoring commenced in theCentre Hills in December 1997 and these data allowa rare glimpse of the response of a bird communityto such a major disturbance event In this paper wereport population changes between 1997 and 2005encompassing four major ashfalls and the inter-vening recovery periods We examine whether ashfallshave caused temporal changes in speciesrsquo abundancesndash both decline and recovery ndash and whether birds

redistribute themselves in the immediate aftermathof ashfalls We also examine the effect of variationsin seasonal rainfall on population trends

METHODS

Study area and study species

Montserrat has a moist tropical climate with rainfallvarying between 1000 and 2500 mm per year Thenatural climax vegetation is a succession from xero-phytic scrub through seasonal forest evergreenrainforest and elfin woodland as precipitation andaltitude increase (Blankenship 1990) There arethree main hill ranges ndash the South Soufriegravere HillsSoufriegravere Hills and Centre Hills rising to 700ndash900 masl Prior to 1995 forests covered most land above

c

250 m in these hill ranges (Fig 1) At lower altitudesthere are narrow strips of riparian forest only

The current eruption has destroyed all of the hillforest in the South Soufriegravere and Soufriegravere Hillswith the exception of a small (

c

200-ha) patch in thesoutheast of the South Soufriegravere Hills Severalnarrow fringes of degraded riparian forest in thelowlands of the north of the island continue to sup-port small populations of some forest bird speciesincluding Forest Thrush Other than this Montserratrsquosforest bird community is now confined to the CentreHills forest (

c

1400 ha) The Centre Hills wereaffected by major ashfalls in AugustndashOctober 1997There was a strong ashfall gradient within theCentre Hills such that sites in the south of the rangereceived ash to a depth of

c

50 mm whereas in thenorth of the range

c

5 mm of ash fell (MontserratVolcano Observatory unpubl data) Another sub-stantial ashfall occurred on 3 July 1998 A thirdmajor ashfall occurred on 29 July 2001 Accuratedata on ash levels are not available but ash depthswere similar to the 1997 events (CGR Bowden perscomm) On 1213 July 2003 a major dome collapsedeposited 150 mm of ash over southwestern parts ofthe Centre Hills decreasing to only 5 mm in thenortheast (Montserrat Volcano Observatory unpubldata) In the intervening periods there were numerousminor ashfalls In addition the large quantity of ashin the soil was frequently re-circulated by windaction during dry periods

During forest bird monitoring in the Centre Hills12 resident forest birds were recorded in sufficientnumbers to be statistically analysed (Table 1)Arendt

et al

(1999) report absolute populationdensities for these species in the Centre Hills based

300

B Dalsgaard

et al


on distance-sampling (Buckland

et al

1993) Thisset includes all of the restricted-range bird speciesknown to occur on Montserrat except Green-throated Carib Antillean Euphonia and LesserAntillean Bullfinch which reach highest densities indrier lowland sites on the island (Montserrat ForestDepartment unpubl data)

Bird census

All bird count data consist of point-counts conductedat 47 fixed points that were visited repeatedly overthe course of the monitoring scheme The points

were taken from a systematic grid of 140 pointsplaced across the entire Centre Hills (see Arendt

et al

1999) All points were gt 300 m apart so theyare independent with respect to detections ofindividual birds For pragmatic reasons the 47 pointssampled from the main grid could not be chosenentirely at random Rather they were chosen tocreate a series of routes that could be walked in asingle day and to cover a representative sample ofhabitats in the Centre Hills In practice this meansthat they are slightly spatially autocorrelated andtend to be in more accessible areas of the forestHowever they give good overall coverage of the

Figure 1 Map of Montserrat showing location of Soufriegravere Hills volcano Soufriegravere Hills South Soufriegravere Hills and the Centre Hills studyarea Shaded areas show approximate outlines of hill forest areas the extant Centre Hills forest the destroyed forest in the Soufriegravereand South Soufriegravere Hills and the surviving forest patch in the South Soufriegravere Hills



301

Centre Hills and the habitats contained thereinincluding the full altitudinal range

Each point-count was conducted for 10 minfollowing a 3-min settling down period Counts wereconducted by two or more observers at least one ofwhom was experienced with the Montserratianavifauna Counts were conducted between 0545and 1500 h and were suspended during heavyrain All bird detections were recorded Full detailsof census techniques are reported in Arendt

et al

(1999) and Hilton

et al

(2003)The 47 points were visited repeatedly beginning

in December 1997 until June 2005 Censuses wereconducted in four seasons DecJan MarApr JunJul and SepOct Most species in Montserrat breedduring MarchndashJune so that the JunJul census includesfledglings from the breeding season just completedwhereas the subsequent MarApr census representsthe pre-breeding situation Because of this we defineddiscrete lsquoyearsrsquo in analysis as running from the JunJul census to the MarApr census During December1997 to November 2001 censuses were conducted

quarterly Following this censuses were conductedless frequently In most censuses a few points weremissed Overall data are available for 25 censuses

Spatial distribution of ashfall

Spatial variation in ashfall was determined fromisopach maps provided by the Montserrat VolcanoObservatory for two of the four major ashfalls(AugustndashOctober 1997 and July 2003) For each ofthese events the isopach maps were used to estimatedepth of ash coverage for each census point and anoverall average ash depth (lsquoAsh-levelrsquo) was derivedand used as a score of the local level of ashing Thespatial pattern of ashfall differed somewhat betweenthe two events During 1997 heaviest ashing was inthe south of the Centre Hills declining toward thenorth In 2003 ashing was heaviest in the southwestdeclining towards the northeast The 1998 and 2001ashfalls were broadly speaking heaviest in the southand lowest in the north but detailed informationis lacking

Table 1 Study species of resident forest bird in the Centre Hills Montserrat

Species Global status Mass (g)dagger DietDagger Feeding zoneDagger

1 Scaly-naped Pigeon Least Concern 288 Frugivoreherbivore Arboreal (terrestrial)Columba squamosa

2 Bridled Quail-dove Restricted-range 230dagger Granivoreinsectivore TerrestrialGeotrygon mystacea

3 Mangrove Cuckoo Least Concern 64 Insectivore UnderstoreyCoccyzus minor

4 Purple-throated Carib Restricted-range 9 Nectarivoreinsectivore Understorey amp canopyEulampis jugularis

5 Antillean Crested Hummingbird Restricted-range 38 Nectarivoreinsectivore Understorey amp canopyOrthorynchus cristatus

6 Caribbean Elaenia Restricted-range 226 Insectivorefrugivore AerialElaenia martinica

7 Forest Thrush Restricted-range 100 Insectivorefrugivore Terrestrial (understorey amp canopy)Cichlherminia lherminieri Vulnerable

8 Scaly-breasted Thrasher Restricted-range 72 Frugivoreinsectivore CanopyMargarops fuscus

9 Pearly-eyed Thrasher Restricted-range 105 Frugivoreinsectivore CanopyMargarops fuscatus carnivore

10 Brown Trembler Restricted-range 53 Insectivore Terrestrial (understorey)Cinclocerthia ruficauda

11 Bananaquit Least Concern 10 NectarivoreInsectivore UnderstoreyCoeroba flava

12 Montserrat Oriole Restricted-range 35 Insectivore UnderstoreyIcterus oberi Critically Endangered

Based on BirdLife International (2004) and Stattersfield et al (1998)daggerMist-net capture data from Montserrat (RSPBMALHE unpubl data) and Dunning (1992)DaggerAuthorsrsquo pers obs Raffaele et al (1998) and del Hoyo et al (1997 1999)

302

B Dalsgaard

et al


Other environmental variables

Four habitat characteristics of each point were meas-ured during August 2003 (Table 2) These variablescapture major axes of variation in the Centre Hillsenvironment particularly vegetation structure andwater availability but are relatively simple to measurePrecipitation data from the Hope Ghaut (southwestCentre Hills 16

deg

45

prime

N 62

deg

13

prime

W) rain gauge wereused to assess the effect of rainfall on populationtrends Data were missing for 1997 and 1998 Forthose years we therefore interpolated values bylinear regression using data available from thenearby island of Nevis (ftpftpncdcnoaagovpubdatainventoriesCOUNTRY-LISTTXT ftpftpncdcnoaagovpubdatainventoriesGLOBAL-SFC-1900-98DTXT) Long-term data indicate anextremely high correlation between Nevis andMontserrat rainfall (1982ndash96 annual totals

r

= 093

P

lt 0001)

Data analysis

Counts for each species were initially modelled asPoisson distributed data with a log link functionWhere data were underdispersed in a Poisson model(deviance

df

lt 09) we modelled the data as a binarydistribution (present or absent) with a binomialerror structure and logit link function (Crawley 1993)Binomial models were used for Bridled Quail-dovePurple-throated Carib Caribbean Elaenia BrownTrembler and Mangrove Cuckoo For these speciesgt 75 of counts were zero and lt 10 of countswere of two or more birds All Poisson modelshad deviance

df

lt 16 and therefore overdispersioneffects were minor

We used PROC GENMOD in the statisticpackage SAS v91 for the analysis In all modelslsquoPointrsquo

times

lsquoSeasonrsquo was declared as a repeated-measures

variable with a first-order autoregressive covariancestructure to account for autocorrelation amongsuccessive counts at the same points while alsocontrolling for seasonal variation in abundance anddetectability All Poisson models were rescaled usingthe DSCALE option Type III Generalised Estimat-ing Equations (GEE) were used to estimate the Wald

χ

2

statistic and hence

P

-value for each explanatoryvariable

Speciesrsquo population trends

We first examined annual variation in bird countsusing lsquoYearrsquo (

n

= 8 years) as a categorical explanatoryvariable lsquoYearrsquo was numbered from the JunJulcensus in calendar year

t

to the MarApr census incalendar year

t

+1

The parameter estimates given foreach year by this model were used as the Populationindex values (see following section)

Effects of ashfall and rainfall on population trends

We next modelled bird counts as a function of ashfalland rainfall while accounting for possible densitydependence in counts For ashfall we modelled foreach species

lsquoCountrsquo

year

t

+1

=

lsquoRecent ashfallrsquo

+

lsquoPopulation indexrsquo

year

t

where lsquoRecent ashfallrsquo is a binary variable which isgiven a value of 1 in censuses that were completedbetween a major ashfall (ie AugustndashOctober 1997July 1998 July 2001 July 2003 see above) and thesubsequent breeding season (commencing in April)and lsquoPopulation indexrsquo is the modelled annualpopulation index (previous section) For rainfallanalyses we replaced lsquoRecent ashfallrsquo with lsquoWetseason rainfallrsquo

year

t

and lsquoDry season rainfallrsquo

year

t

Table 2 Habitat variables measured at each point on the census grid

Variable Variable type Method of measurement

Distance to watercourse

Continuous (m) Assigned to one of four distance bands 0ndash25 m 25ndash50 m 50ndash100 m and gt 100 m with the midpoint of each distance band being used to develop a continuous variable

Visually estimated in the field by forest rangers who knowthe area well

Rainfall Continuous (mm yrminus1) Derived from isohyet map of Montserrat for the period1935ndash64

Altitude Continuous (m asl) Hand-held altimeterCanopy cover Continuous (arcsine transformed percentage) Estimates were made vertically through a hand-held tube

10 times along a 20-m transect through the census point



303

We separately analysed wet and dry season rainfallbecause we

a priori

expected them to have differingand perhaps opposing effects on bird counts Thewet season in Montserrat is the hurricane season(JulyndashDecember) and is prone to periods of coolweather strong winds and heavy rain whereas thedry season (JanuaryndashJune) is prone to occasionalperiods of drought The bulk of the birdsrsquo breedingseason occurs during the dry season The effects ofrainfall are expected to arise primarily throughmedium-term impacts on forest ecology (eg plantgrowth and flowering insect abundance) Hencewe modelled the effect of rainfall in the study yearprior to the counts Conversely ashfalls are relativelyinstantaneous events the effects of which (coveringsurfaces with ash) are likely to be initially large andthen to diminish rapidly Therefore we modelledashfall effects on subsequent counts in the samestudy year

Spatial and temporal effects of ash on speciesrsquo abundances

To assess the effect of ashfall on counts in moredetail models were developed for each species usingboth spatial and temporal variation in ashfall asexplanatory variables (and again accounting fordensity dependence)

lsquoCountrsquo

year

t

+1

=


+

lsquoAsh-levelrsquo + lsquoRecent ashfallrsquo

times

lsquoAsh-levelrsquo

+


year

t

where lsquoAsh-levelrsquo reflects spatial variation in thequantity of ash falling across the Centre Hills overthe entire course of the eruption (see above) Hencethis model evaluates (1) whether counts vary inresponse to recent ash deposition in the Centre Hills(significant effect of lsquoRecent ashfallrsquo) (2) whethercounts over the whole study period vary in responseto spatial variations within the Centre Hills in theamount of ash falling (significant effect of lsquoAsh-levelrsquo) and (3) whether the temporal response toashfall varies between high-ash and low-ash areas(significant effect of lsquoRecent ashfallrsquo

times

lsquoAsh-levelrsquo)As spatial variation in the amount of ash may be

confounded by variation in other habitat featureswe also analysed counts in relation to lsquoAsh-levelrsquoafter controlling for habitat associations We developedhabitat association models for each species using theexplanatory habitat variables described in Table 2Quadratic altitude canopy cover and rainfall effects

were included in the initial models but no inter-action terms were permitted

t

-tests indicated thataltitude and rainfall differ significantly between lowand high ash points (low ash points being loweraltitude and higher rainfall) but that values of otherhabitat variables did not differ significantly betweenthe two ash-level categories

Full habitat association models were developedwith subsequent step-wise deletion of non-significantinteraction terms and then main effects until aminimal adequate model (MAM) with all variablessignificant (P lt 005) was obtained A categoricallsquoYearrsquo term was included in all models to account fortemporal variation in counts lsquoPointrsquo times lsquoSeasonrsquo wasas in all models declared as a repeated-measuressubject The normally distributed standardizedresiduals from the MAM were then used as aresponse variable with lsquoAsh-levelrsquo as an explanatoryvariable in a GEE model (with lsquoPointrsquo times lsquoSeasonrsquo as arepeated measure) to test for significant effects oflsquoAsh-levelrsquo on bird counts

RESULTS


Species trends across eight years showed a ratherhigh degree of concordance Most species increasedduring 19978ndash20001 and almost all speciesdecreased in abundance between 20012 and 20023(Fig 2) Trends subsequent to 2003 were varied Incontrast to the general pattern Caribbean Elaeniaand Montserrat Oriole decreased between 1997 and2001 Bananaquit showed a remarkably smoothtrend followed by a sharp drop between 20023 and20034 and Purple-throated Carib and CaribbeanElaenia doubled in abundance between 20012 and20023 Figure 3 shows that the mean populationmultiplication rate (λ) tended to vary systematicallyby year among the study species being high initiallyand then declining gradually to a low point in theinterval 20012ndash20023 with a subsequent increasein the final years of the study period Mean λ acrossall species exceeded 1 in all time-intervals except for20001ndash20012 and 20012ndash20023


Given that there is a common temporal pattern inλ among most species (Fig 3) does this relate totemporal environmental variation Four species

304 B Dalsgaard et al


ndash Scaly-naped Pigeon Bridled Quail-dove ForestThrush and Brown Trembler ndash showed strong andsignificant declines in response to lsquoRecent ashfallrsquo(Table 3) Perversely Mangrove Cuckoo appeared toincrease substantially following ashfalls No effect ofrecent ashfall was detected in counts of the remainingseven species Hence for a significant part but not amajority of the forest bird community there was

evidence that ashfalls induce relatively sharp short-term declines in counts

However the models suggested that rainfall inthe previous year had an even stronger impact onbird counts Most species abundances respondedsignificantly to rainfall in the previous year (Table 4)A group of species (Scaly-naped Pigeon BridledQuail-dove Purple-throated Carib Forest Thrush

Figure 2 Population trends of resident forest birds in the Centre Hills Montserrat 1997ndash2005 Points and solid lines fitted annualestimates from Generalised Estimating Equations (GEEs) using lsquoYearrsquo as a categorical factor plusmn 95 confidence limits Values are back-transformed parameter estimates from GEEs Hence for Poisson models (Scaly-naped Pigeon Antillean Crested Hummingbird ForestThrush Scaly-breasted Thrasher Pearly-eyed Thrasher Bananaquit and Montserrat Oriole) values represent estimated count per pointFor binomial models (Bridled Quail-dove Purple-throated Carib Caribbean Elaenia Brown Trembler and Mangrove Cuckoo) valuesrepresent estimated probability of detecting the species per point


Volcanic impacts on forest birds 305

and Pearly-eyed Thrasher) responded positively towet-season rainfall by increasing in abundance inthe subsequent year None of these species showedthe same positive response to dry-season rainfall Scaly-naped Pigeon and Pearly-eyed Thrasher decreased

following wet dry seasons Forest Thrush showed anear-significant trend in the same direction whilePurple-throated Carib and Bridled Quail-doveshowed no response to dry-season rainfall AntilleanCrested Hummingbird and Brown Trembler both

Figure 3 Population multiplication rate of 12 forest bird species in relation to rainfall and ashfall in Montserratrsquos Centre Hills 1997ndash2005 Small dots values of population multiplication rate (λ = population indexyear t +1population indexyear t) for individual species in eachyear interval Thick continuous line geometric mean value of λ large open circles wet-season rainfall in yeart large closed circles dry-season rainfall in yeart vertical arrows time intervals in which a major ashfall occurred in yeart

Table 3 Effect of recent volcanic ashfall on population trends of forest birds

Species

Recent ashfall

Effect ( change) Wald χ2 (df = 1) P

Scaly-naped Pigeon minusminusminusminus51 229 lt 00001Bridled Quail-dove minusminusminusminus32 409 0043Mangrove Cuckoo +30 455 0033Purple-throated Carib minus13 137 024Antillean Crested Hummingbird minus11 184 017Caribbean Elaenia +12 089 035Forest Thrush minusminusminusminus27 128 00003Scaly-breasted Thrasher minus39 025 061Pearly-eyed Thrasher minus86 288 0090Brown Trembler minusminusminusminus28 386 0049Bananaquit +071 001 090Montserrat Oriole +13 085 036

Based on Generalised Estimating Equations with lsquoRecent ashfallrsquo as a binary variable lsquoYearrsquo as a fixed effect and lsquoPointrsquo times lsquoSeasonrsquo asa repeated-measures effectRows in bold are significant effects (P lt 005)Modelled percentage difference in counts between a census conducted following a recent ashfall and censuses conducted in theabsence of recent ashfall Positive values indicate that counts increased following ashfalls and vice versa



showed the same positive responses to wet-seasonrain and negative responses to dry-season rain butwithout such strong statistical support

In contrast to this pattern across seven of the studyspecies Caribbean Elaenia increased following dryseasons but showed no response to wet-season rain-fall Mangrove Cuckoo showed a different responsealtogether it increased following years of low rainfallin wet and dry seasons The remaining species ndashScaly-breasted Thrasher Bananaquit and MontserratOriole ndash showed no suggestion of a populationresponse to rainfall

In full models including both rainfall and ashfallvariables simultaneously the idea that rainfall wasthe more important determinant of populationtrends was borne out (Table 5) However these fullmodels suggested that dry-season rather than wet-season rainfall may be the more important factorFive species showed a negative trend in response todry-season rainfall ie they tended to increase whenthe previous dry season was relatively dry For BrownTrembler there was a similar marginally non-significant effect Caribbean Elaenia showed a reversepattern increasing after wet dry seasons Threespecies (Purple-throated Carib Forest Thrush andPearly-eyed Thrasher) responded positively towet-season rainfall in the full models

Conversely the effects of recent ashfall tendedto weaken when rainfall effects were controlled for

in the full models There was evidence of strongnegative effects of ash on counts of Scaly-napedPigeon and Brown Trembler only For AntilleanCrested Hummingbird there was a marginallynon-significant negative effect of ash For BridledQuail-dove and Forest Thrush the negative effectsof ash that were apparent from the ash-only modelsdisappeared in the full models The perverse positiveeffect of ash on Mangrove Cuckoos disappeared whenrainfall was accounted for


The outcomes of models analysing the effect ofrecent major ashfall spatial variation in ash levelsand the interaction between them are shown inTable 6 These models suggest that the effects of ashwere relatively slight and not widespread across theavifauna Temporal effects were apparent only in thetwo Columbid species which were significantly lessabundant in counts conducted after recent ashfallsSpatial effects were found in Montserrat Oriolewhich tended to be less abundant in areas with highashfall Conversely Mangrove Cuckoo was signifi-cantly more abundant in high ash areas For ForestThrush alone there was evidence of a redistributionaway from high ash areas in immediate response toashfalls

Table 4 Effect of seasonal rainfall on population trends of forest birds

Wet-season rainfall Dry-season rainfall

SpeciesEffect

( change)Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Scaly-naped Pigeon +15 547 0019 minusminusminusminus19 261 lt 00001Bridled Quail-dove +25 610 0014 +23 012 073Mangrove Cuckoo minusminusminusminus22 925 00024 minusminusminusminus15 887 00029Purple-throated Carib +20 195 00021 +66 155 021Antillean Crested Hummingbird +62 178 018 minusminusminusminus81 533 0021Caribbean Elaenia minus12 238 012 +18 900 00027Forest Thrush +29 394 lt 00001 minus56 282 0093Scaly-breasted Thrasher +090 003 087 +26 064 042Pearly-eyed Thrasher +11 207 lt 00001 minusminusminusminus12 492 lt 00001Brown Trembler +14 219 014 minus14 318 0074Bananaquit +30 079 037 minus30 078 038Montserrat Oriole minus12 002 088 +39 031 058

Based on Generalised Estimating Equations with lsquoWet-season rainfallrsquo or lsquoDry-season rainfallrsquo in the previous years as a covariate lsquoYearrsquoas a fixed effect and lsquoPointrsquo times lsquoSeasonrsquo as a repeated-measures effectRows in bold are significant effects (P lt 005)Modelled percentage change in counts if rainfall is 1 sd above the mean relative to abundance if rainfall takes the mean value(n = 8 years of rainfall data) Positive values indicate that counts increased following high rainfall and vice versa



The models in which habitat associations werecontrolled for also gave little evidence for avoidanceof high ash areas by forest birds Only one speciesBridled Quail-dove showed a significantly lowerabundance in high ash areas after habitat associationswere controlled for (Table 7)

DISCUSSION

These analyses provide some of the first informationconcerning the effect of the Soufriegravere Hills volcaniceruption on Montserratrsquos bird community It alsogives a unique insight into the response of avifaunas

Table 5 Effect of seasonal rainfall and ashfall on population trends of forest birds

Species

Wet-season rainfall Dry-season rainfall Ashfall

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Scaly-naped Pigeon minus24 384 lt 00001 ndash ndash ndash minus56 298 lt 00001Bridled Quail-dove ndash ndash ndash ndash ndash ndash ndash ndash ndashMangrove Cuckoo minus11 405 0044 minus18 298 0084 ndash ndash ndashPurple-throated Carib ndash ndash ndash +26 111 0009 ndash ndash ndashAntillean Crested Hummingbird minus13 130 00003 ndash ndash ndash minus19 347 0063Caribbean Elaenia +22 158 lt 00001 ndash ndash ndash ndash ndash ndashForest Thrush minus14 158 lt 00001 +30 188 lt 00001Scaly-breasted Thrasher ndash ndash ndash ndash ndash ndash ndash ndash ndashPearly-eyed Thrasher minus14 522 lt 00001 +14 222 lt 00001Brown Trembler minus23 776 0054 ndash ndash ndash minus39 487 0027Bananaquit ndash ndash ndash ndash ndash ndash ndash ndash ndashMontserrat Oriole ndash ndash ndash ndash ndash ndash ndash ndash ndash

Based on Generalised Estimating Equations with lsquoWet-season rainfallrsquo lsquoDry-season rainfallrsquo and lsquoRecent ashfallrsquo as explanatoryvariables lsquoYearrsquo as a fixed effect and lsquoPointrsquo times lsquoSeasonrsquo as a repeated-measures effect Non-significant explanatory variables weresequentially deleted All effects with P lt 010 are shown in the tableModelled percentage change in counts if rainfall is 1 sd above the mean relative to abundance if rainfall takes the mean value(n = 8 years of rainfall data) Positive values indicate that counts increased following high rainfall and vice versa

Table 6 Effects of spatial and temporal variation in volcanic ash forest bird counts

Recent ashfall Ash-level lsquoRecent ashfallrsquo times lsquoAsh-levelrsquo

SpeciesDirection of effect P

Direction of effect P


Scaly-naped Pigeon ndash 00003 + 0065 + 082Bridled Quail-dove ndash 0035 ndash 0066 + 058Mangrove Cuckoo + 027 + lt 00001 + 049Purple-throated Carib ndash 071 ndash 050 ndash 042Antillean Crested Hummingbird ndash 059 ndash 037 ndash 051Caribbean Elaenia + 065 + 068 + 069Forest Thrush ndash 019 + 0055 ndash 0043Scaly-breasted Thrasher + 095 + 032 ndash 041Pearly-eyed Thrasher ndash 061 ndash 089 ndash 0087Brown Trembler ndash 012 + 020 + 087Bananaquit ndash 042 + 024 + 0062Montserrat Oriole + 074 ndash 0025 ndash 050

Based on Generalised Estimating Equations with lsquoRecent ashfallrsquo lsquoAsh-levelrsquo and lsquoRecent ashfallrsquo times lsquoAsh-levelrsquo as explanatory variableslsquoYearrsquo as a fixed effect and lsquoPointrsquo times lsquoSeasonrsquo as a repeated-measures effectP-values marked in bold are significant (P lt 005)A negative effect implies that counts in areas experiencing high ashfalls were relatively lower during censuses when there had beenlsquoRecent ashfallrsquo and vice versa



to volcanic ashfalls The generality of the responsesobserved in Montserrat is unclear Volcanic erup-tions in different parts of the world vary in the quan-tity and nature of ash falling and its persistence inthe environment The bird community sampled herecontains a mix of forest insectivores omnivoresnectarivores and frugivores although being from asmall island that has a history of catastrophic eco-logical disturbance it is characterized by generalistspecies which may be more adaptable to disturbancethan some other avifaunas

The loss of c 60 of the forested land was coupledwith massive disruption of the remaining forest inthe Centre Hills Major ashfalls resulted in large-scaledefoliation and anecdotal reports from 19978 de-scribe insect die-offs within the forest Major ashfallssubstantially depress canopy arthropod numbers(Marske 2004) Despite this the Montserratian birdcommunity has remained qualitatively unalteredno resident forest bird species have been extirpatedFurthermore the relative abundance of species hasnot changed greatly (Fig 2) By contrast majorhurricanes are thought to have caused extremepopulation crashes on other Caribbean islands inparticular of nectarivoresfrugivores (eg Wauer ampWunderle 1992 Rathcke 2000)

The volcano shifted from a period of intenseactivity and very frequent ashfalls in 1997 to a period

of relative quiet interspersed with occasional (lessthan annual) major ashfalls during 1998ndash2001During this time the majority of forest bird speciesshowed very large-scale population increases (Fig 2)Superficially this suggests dramatic quantitativeeffects of ashfall on the abundance of most birdspecies in the Centre Hills and the 1997ndash2001 increasescan be interpreted as a recovery phase followingthe peak of volcanic activity Furthermore the patternfor several species was for very substantially lowercounts in censuses following ashfalls (Table 3)suggesting an immediate acute effect of major ashfallson bird populations However for eight of the 12study species there was no evidence of a reductionin counts immediately following an ashfall Inaddition when the influence of seasonal rainfall wasexamined it appeared to explain a substantial partof the variation in bird numbers and to reduce theapparent importance of ashfalls (Tables 4 amp 5)Finally the large variation in the quantity of ashfalling in different parts of the Centre Hills shouldgenerate strong spatial variation in the observed birdpopulation responses and yet there was little evidencefor reduced counts in areas that were most affectedby ash nor of redistribution of birds in response toashfall (Tables 6 amp 7) Redistribution of BullockrsquosOrioles Icterus bullockii away from high ash areasfollowing ashfalls was observed at Mount St Helens

Table 7 Effect of spatial variation in volcanic ash on forest bird counts after controlling for habitat associations

Species

Habitat variables in modellsquoAsh-levelrsquo parameter estimatedagger

Wald χ2 (df = 1) PAltitude Rainfall

Canopy cover


Scaly-naped Pigeon ndash 00016 063 043Bridled Quail-dove minusminusminusminus00027 606 0014Mangrove Cuckoo n + + + 00014 083 036Purple-throated Carib n ndash minus00001 001 091Antillean Crested Hummingbird + ndash n ndash 00017 232 013Caribbean Elaenia ndash + minus00003 003 087Forest Thrush ndash ndash 00016 065 042Scaly-breasted Thrasher ndash 00005 009 076Pearly-eyed Thrasher ndash minus00005 006 080Brown Trembler 00010 050 048Bananaquit + n ndash ndash 00000 000 099Montserrat Oriole n minus00024 232 013

Habitat association models were developed using Generalised Estimating Equations with habitat variables (linear and quadratic) andlsquoYearrsquo as explanatory variables and lsquoPointrsquo times lsquoSeasonrsquo as a repeated-measures effect Standardized residuals from these models werethen used as response variables in General Linear Models with lsquoAsh-levelrsquo as a continuous explanatory variable Values in bold indicatesignificant effects (P lt 005) of spatial variation in volcanic ash on forest bird counts after controlling for habitat associationsSymbols used to describe the relationship between bird abundance and habitat variables + positive association ndash negativeassociation n quadratic association with peak bird abundance at intermediate values of the explanatory variabledaggerPositive values indicate that counts were higher at points with relatively high ash levels and vice versa



(Butcher 1981) We therefore suggest that althoughthere were reductions in populations of some speciesimmediately following ashfalls these were relativelyshort-lived and several species seemed remarkablyunaffected

The proximate causes of the low counts followingashfalls are not clear Possibly changes in the birdsrsquobehaviour might make them less conspicuous toobservers (eg if stress and food shortage made themless vocal) However this seems rather unlikelyin the immediate aftermath of ashfalls birds areparticularly visible because of the pale-coloured ashbackground and the extensive defoliation

Nest failure resulting from abandonment andcovering of eggs with ash was recorded in BullockrsquosOrioles and California Gulls Larus californicusfollowing the Mount St Helens eruption (Butcher1981 Hayward et al 1982) Andersen and MacMahon(1986) showed that predation rates were extremelyhigh in ashfall areas During major ashfalls onMontserrat active nests of Montserrat Orioles weredirectly destroyed by falling ash (RSPBMALHEunpubl data) and this probably applies to all otheropen-nesting species However the major ashfallsin Montserrat occurred late in or shortly after thebreeding season so breeding failure seems relativelyunlikely as a cause of population declines

As the major ashfalls occurred near the start of thepost-breeding period low counts in the censuses thatoccurred between the ashfalls and the subsequentbreeding season imply that temporarily increasedmortality of adult and fledged juveniles was theproximate cause of the declines

Reductions in arthropod numbers following ash-falls appear to be ephemeral but pronounced (Marske2004) which may cause temporary food shortagesAt Mount St Helens there was also evidence ofmarkedly lower insect abundance shortly after theeruption (Foster amp Myers 1982) as ash from theeruption was lethal to insects (Edwards amp Schwartz1981) Following ashfalls birds may also be moreconspicuous to predators ndash in this instance primarilyAmerican Kestrels Falco sparverius Merlins Falcocolumbarius and Pearly-eyed Thrashers (Arendt 2006)Hayward et al (1982) showed that conjunctivitiscan occur in birds following ashfall It has also beenshown that exposure to ash affects the human respi-ratory system (Searl et al 2002) However duringmist-netting of c 400 birds in the Centre Hills during1997ndash2005 no externally obvious signs of ill-healthwere observed among Montserratrsquos avifauna Immi-gration of large numbers of birds into the Centre Hills

from the southern hill ranges might have occurredas these latter forests were destroyed however thepattern of population changes in the Centre Hillsstrongly suggests that this was not the case Thesouthern forests were effectively destroyed by late1997 while forest bird populations in the Centre Hillsat that time were apparently at their lowest (Fig 2)

The apparently rapid population recovery followingashfall may indicate density-dependent improvementsin demographic rates following ash-induced popu-lation decreases Alternatively the forest may behighly productive following an ashfall due to theinput of nutrients from ash and a flush of new plantgrowth following defoliation In either scenario therecovery is likely to be driven by relatively highreproductive success in the breeding season followingan ashfall The net result is that following an 8-yearperiod in which there were four major ashfalls mostspecies were at least as abundant as at the outsetSimilarly at Mount St Helens songbird populationshaving been depressed in the year of the eruptionwere largely back to pre-eruption levels in thesubsequent year (Foster amp Myers 1982) Manuwalet al (1987) who studied community-level effects ofthe Mount St Helens eruption 2ndash5 years after theevent found that at a site that was lightly ashed birdcommunities were little different from those in similarhabitats in unaffected areas At a second site whichsuffered heavy vegetation loss through scorchingthe numbers of tree-foliage insectivores and tree seed-eaters were dramatically reduced relative to ground-foraging birds We were unable to study forestbird communities in scorched areas of Montserratbecause such areas occur only in the lsquoexclusion zonersquo inthe south of the island where the volcanorsquos continuedactivity makes access impossible

Our data suggest that seasonal variation in rainfallplays a strong role in the population dynamics ofMontserratrsquos forest birds with the majority of speciesshowing a significant response to either dry-seasonrainfall or wet-season rainfall The most frequentresponse to rainfall was for species to decrease ifthe preceding dry season was relatively wet and toincrease if the preceding wet season was relativelywet This implies that dry breeding seasons (dry sea-son) resulted in high fecundity andor that survivalof fledglings and adults was high in wet non-breedingseasons Anecdotal observations suggest that theformer is unlikely as a general response a dry-seasondrought in 2001 caused widespread breeding deferraland failure in the Centre Hills (RSPBMALHEunpubl data)



A complication arises because during the studyyears there was a tendency for wet-season rainfall(JulyndashDecember) to be positively correlated withdry-season rainfall in the following year (JanuaryndashJune) (Fig 3) Hence it is rather difficult to discernwhich season is the key driver of population trendFurthermore there was a tendency for ash-years tofollow dry years (Fig 3) so again the effects of ashfalland rainfall are difficult to separate Thus althoughwe have some evidence that rainfall may be animportant driver of forest bird population dynamicsin Montserrat more data are needed before theeffect can be separated from that of ashfall and itsexact form can be determined

It is difficult to discern whether any ecologicaltraits predispose species to be sensitive to ashfallThe species that show evidence of declines andorredistributions away from ashfalls are Scaly-napedPigeon Bridled Quail-dove Antillean CrestedHummingbird Forest Thrush Brown Trembler andMontserrat Oriole (although the evidence is equivocalfor some of these) Assigning Lesser Antillean birdspecies to feeding guilds is problematic because theyare almost all generalists feeding on more than onemajor food type (Table 1) This in itself may be aresponse to the high level of environmental stochas-ticity in Lesser Antillean habitats which militatesagainst specialization Although one food type maybe dominant under normal conditions diet maychange in response to stressful conditions such asashfalls for example Montserrat Orioles have beenobserved feeding on fruit during ashfall episodes butvery rarely at other times (RSPBMALHE unpublobs) Butcher (1981) suggested that in generalsmall insectivorous birds were most affected by theash from the Mount St Helens eruption but thereis no evidence for this in the Montserrat forest birdcommunity and indeed no particular diet typeappears to be more sensitive than others nor doesbody size seem to predict sensitivity There arehowever signs of a link between primary feedingzone and sensitivity all of the terrestrially foragingspecies show evidence of declines or redistributionin response to ashfall This is in agreement withMarskersquos (2004) suggestion that canopy arthropodsare less severely affected by ashfall than terrestrialarthropods because ash is removed from the canopyrelatively quickly by rainfall and leaf-fall whereas itremains in the soil far longer

Overall the effects of the eruption are perhapsless severe than might be anticipated for residentbirds on such a small island These results are in

contrast to data concerning the fruit-bat communityin Montserrat which has been profoundly alteredTwo of the nine fruit-bat species present on theisland in the mid-1990s are apparently extirpatedand capture rates of all species declined greatly in19978 (Adams amp Pedersen 1999 Pedersen 2000)Data concerning the impact of the eruption on theterrestrial reptiles and amphibians of Montserrathave not been published

One might expect the massive ash deposition tocause longer-term changes to the ecology of the CentreHills Volcanic ash is known to change soil character-istics (Nammah et al 1986) to reduce growth of plantseedlings (Caruso et al 1990) and to stress especiallysensitive plants (Sadler amp Grattan 1999) The effectsof acid rain on forest plants and decomposers are stilldebated (eg Bormann 1985 Larkin amp Kelly 1987Chen et al 1991 Sherman amp Fahey 1994 Chang ampLee 1995 Heneghan amp Bolger 1996) One mightalso predict indirect effects on bird communitiesThe introduced rat population in the Centre Hillsincreased dramatically from 1998 to 2001 (bothBlack Rat Rattus rattus and Brown Rat R norvegicusare abundant although whether one or both speciesincreased is not known) before subsequently decliningIt is unclear whether this was in response to volcanicdisturbance but it is known that nest predation byrats on Montserrat Orioles and hence presumablyother sensitive bird species was extremely high in2001 However such long-term and indirect effectswill be superimposed on an ecosystem that is alreadysubstantially influenced by human activity The forestis largely secondary following clearance in the plan-tation era (Beard 1949) and still contains numerousnon-native fruit trees Introduced Pigs Sus scrofa andrats are widespread Agricultural and built areas fringethe hills with consequent benefits for species thatthrive in anthropogenic landscapes such as Pearly-eyed Thrasher a major nest-predator (Arendt 2006)The effect of the eruption might be surprisinglysmall relative to other external influences on the forest

If the current eruption continues one can expectfurther temporary reductions in Montserratianforest bird populations However as long as theashfalls are intermittent and are not compoundedby other catastrophes forest bird populations appearable to persist After 9 years of the current eruptionMontserrat remains a major site for many of therestricted-range species of the Lesser Antilles EBA

This study was funded by the UK Foreign amp Common-wealth Office American Bird Conservancy WWFNovo



Nordisk and University of Aarhus We are grateful to TonyHill (DFID) Richard Allcorn Chris Bowden Ken SmithMark Hulme Joah Madden Liz Mackley and Debbie Pain(RSPB) and Phil Atkinson (British Trust for Ornithology)for valuable discussion and fieldwork assistance and toDavid Buckingham Richard Bradbury and Paul Donald(RSPB) and Preben Blaeligsild (University of Aarhus) forstatistical suggestions David Robinson and John Guilbertkindly provided Nevis rainfall data The MontserratVolcano Observatory (MVO) provided ashfall data forthe Centre Hills Professor Mike Ivie and Katie Marske(Montana State University) provided valuable additionaldata and discussion The manuscript was greatly improvedby the suggestions of Gavin Siriwardena and an anony-mous referee

REFERENCES

Adams R amp Pedersen S 1999 The effects of natural disas-ters on bat populations on Montserrat BWI a 20 year historyAm Zool 38 52

Andersen DC amp MacMahon JA 1986 An assessment ofground-nest depredation in a catastrophically disturbedregion Mount St Helens Washington Auk 103 622ndash626

Arendt WJ 2006 Adaptations of an Avian Supertramp Distri-bution Ecology and Life History of the Pearly-eyed Thrasher( Margarops fuscatus) General Technical Report 27 SanJuan PR US Department of Agriculture Forest ServiceInternational Institute of Tropical Forestry

Arendt WJ Gibbons DW amp Gray G 1999 Status of thevolcanically threatened Montserrat Oriole Icterus oberi andother forest birds in Montserrat West Indies Bird ConservInt 9 351ndash372

Beard JS 1949 The natural vegetation of the Windward andLeeward Islands Oxford Forry Memoirs 21 1ndash192

BirdLife International 2004 Threatened Birds of the World2004 CD-ROM Cambridge UK BirdLife International

Blankenship JR 1990 The Wildlife of Montserrat Including anAnnotated Bird List Montserrat West Indies MontserratNational Trust

Bormann FH 1985 Air pollution and forests ndash an ecosystemperspective Bioscience 35 434ndash441

Bralower TJ Thomas DJ Zachos JC Hirschmann MMRohl U Sigurdsson H Thomas E amp Whitney DL1997 High-resolution records of the late Paleocene thermalmaximum and circum-Caribbean volcanism is there a causallink Geology 25 963ndash966

Buckland ST Anderson DR Burnham KP amp Laake JL1993 Distance Sampling Estimating Abundance of BiologicalPopulations London Chapman amp Hall

Butcher GS 1981 General notes Murrelet 62 15ndash16Caruso JL Sumidastevens S Sovonickdunford S

Winget GD amp Jayasimhulu K 1990 Influence of MountSt-Helens volcanic ash on growth and abscisic-acid inDouglas-fir seedlings J Plant Physiol 137 229ndash232

Chang KS amp Lee SW 1995 Sensitivity and self-purificationfunction of forest ecosystem to acid precipitation (II) Ionbalance in vegetation and soil leachate J Korean For Soc84 103ndash113

Chen C Zhang J Xu G Zhou C amp Liao L 1991 Effectsof acid rain on forest ecosystem in southwestern ChinaJ Environ Sci 3 27ndash36

Crawley MJ 1993 GLIM for Ecologists Oxford BlackwellDunning JC 1992 The CRC Handbook of Avian Body

Masses London CRC PressEdwards JS amp Schwartz LM 1981 Mount St Helens ash

a natural insecticide Can J Zool 59 714ndash715Foster JH amp Myers WL Jr 1982 Ashfall effects on Columbia

basin wildlife Proceedings of the conference Mount StHelens effects on water resources Report No 41 151ndash162

Hayward JL Miller DE amp Hill CR 1982 Mount St Helensash ndash its impact on breeding Ring-billed and California GullsAuk 99 623ndash631

Heneghan L amp Bolger T 1996 Effect of components of lsquoacidrainrsquo on the contribution of soil microarthropods to ecosystemfunction J Appl Ecol 33 1329ndash1344

Hilton GM Atkinson PW Gray GAL Arendt WJ ampGibbons DW 2003 Rapid decline of the volcanicallythreatened Montserrat Oriole Biol Conserv 111 79ndash89

Hilton GM Bowden CGR Ratcliffe N Lucking V ampBrindley E 2001 Bird Conservation Priorities in the UKOverseas Territories RSPB Research Report no 1 SandyUK RSPB

del Hoyo J Elliott A amp Sargatal J (eds) 1997 Handbookof the Birds of the World Vol 4 Sandgrouse to CuckoosBarcelona Lynx Edicions

del Hoyo J Elliott A amp Sargatal J (eds) 1999 Handbook ofthe Birds of the World Vol 5 Barn-owls to HummingbirdsBarcelona Lynx Edicions

Larkin RP amp Kelly JM 1987 Influence of elevated ecosys-tem sulfur levels on litter decomposition and mineralizationWater Air Soil Poll 34 415ndash428

Manuwal D Hugg M Bauer M Chappell C amp Hegstad K1987 Summer birds of the upper subalpine zone of MountAdams Mount Rainier and Mount St Helens WashingtonNorthwest Sci 61 82ndash91

Marske KA 2004 Effects of Volcanic Ash on the Insect Foodof the Montserrat Oriole Icterus oberi Lawrence 1880Masterrsquos thesis Montana State University Bozeman USA

Myers N Mittermeier RA Mittermeier CG da FonsecaGAB amp Kent J 2000 Biodiversity hotspots for conservationpriorities Nature 403 853ndash858

Nammah H Larsen FE McCool DK Fritts R Jr amp Molnau M1986 Mount St Helens volcanic ash effect of incorpo-rated and unincorporated ash of two particle sizes onrunoff and erosion Agriculture Ecosystems Environ 15 63ndash72

Pedersen SP 2000 The impact of volcanic eruptions on thebat populations of Montserrat BWI Am Zool 40 1167ndash1168

Raffaele H Wiley J Garrido O Keith A amp Raffaele J1998 Birds of the West Indies London Christopher Helm

Rathcke BJ 2000 Hurricane causes resource and pollinationlimitation of fruit-set in a bird-pollinated shrub Ecology 811951ndash1958

Sadler JP amp Grattan JP 1999 Volcanoes as agents of pastenvironmental change Global Planet Change 21 181ndash196

Searl A Nicholl A amp Baxter PJ 2002 Assessment of theexposure of islanders to ash from the Soufriegravere Hills volcanoMontserrat British West Indies Occup Environ Med 59523ndash531



Sherman RE amp Fahey TJ 1994 The effects of acid deposi-tion on the biogeochemical cycles of major nutrients in mini-ature red spruce ecosystems Biogeochemistry 24 85ndash114

Stattersfield AJ Crosby MJ Long AJ amp Wege DC1998 Endemic Bird Areas of the World Cambridge UKBirdlife International

Wauer RH amp Wunderle JM Jr 1992 The effect of hurricaneHugo on bird populations on St Croix US Virgin IslandsWilson Bull 104 656ndash673

Wunderle JM 1995 Responses of bird populations in aPuerto-Rican forest to Hurricane Hugo ndash the first 18 monthsCondor 97 879ndash896

Wunderle JM Lodge DJ amp Waide RB 1992 Short-termeffects of Hurricane Gilbert on bird populations on JamaicaAuk 109 148ndash166

Received 11 April 2006 revision accepted 11 August 2006



299

available comes from the eruption of Mount StHelens USA on 18 May 1980 (Hayward

et al

1982Andersen amp MacMahon 1986 Manuwal

et al

1987)Such information is relevant to the conservation of 15globally threatened bird species that are threatenedby volcanic activity and in particular the two birdspecies ndash Black-breasted Puffleg

Eriocnemis nigrivestris

and Mount Karthala White-eye

Zosterops mouroniensis

ndash that have their entire ranges on the slopes of currentlyactive volcanoes (BirdLife International 2004)

Here we examine the influence of the eruptionof the Soufriegravere Hills volcano on the forest birdcommunity in Montserrat (16

deg

N 62

deg

W 109 km

2

) aUK Overseas Territory at the northern end of theLesser Antilles One would predict that on smallislands volcanic eruptions might result in temporarypopulation bottlenecks and possibly extirpationwith recolonization occurring once the volcano hasfallen dormant The current eruption of the SoufriegravereHills volcano began in 1995 and continues at thetime of writing The volcano also erupted in 1640though information about this eruption is scantBriefly the eruption has been characterized by thegrowth of an andesite lava dome with associatedpyroclastic flows vulcanian explosion and debrisflows (wwwmvoms) Pyroclastic flows caused thealmost complete destruction of tropical forest inthe southern hill ranges of the island by late 1997In the remaining forested hill range the Centre Hillsashfalls and acid rains have been frequent andheavy during 1996ndash2005 interspersed with periodsof recovery This pattern of localized obliterationof ecological communities accompanied by muchmore widespread impacts of ashacid rainfall haspresumably been the pattern in previous eruptionsalthough direct evidence is lacking Montserratrsquosavifauna is of considerable conservation importanceas it includes the critically endangered single-islandendemic Montserrat Oriole (Hilton

et al

2003)the vulnerable four-island endemic Forest Thrushand nine other restricted-range species (Stattersfield

et al

1998 Arendt

et al

1999 Hilton

et al

2001)(see Table 1 for scientific names of the study species)

Regular forest bird monitoring commenced in theCentre Hills in December 1997 and these data allowa rare glimpse of the response of a bird communityto such a major disturbance event In this paper wereport population changes between 1997 and 2005encompassing four major ashfalls and the inter-vening recovery periods We examine whether ashfallshave caused temporal changes in speciesrsquo abundancesndash both decline and recovery ndash and whether birds

redistribute themselves in the immediate aftermathof ashfalls We also examine the effect of variationsin seasonal rainfall on population trends

METHODS

Study area and study species

Montserrat has a moist tropical climate with rainfallvarying between 1000 and 2500 mm per year Thenatural climax vegetation is a succession from xero-phytic scrub through seasonal forest evergreenrainforest and elfin woodland as precipitation andaltitude increase (Blankenship 1990) There arethree main hill ranges ndash the South Soufriegravere HillsSoufriegravere Hills and Centre Hills rising to 700ndash900 masl Prior to 1995 forests covered most land above

c

250 m in these hill ranges (Fig 1) At lower altitudesthere are narrow strips of riparian forest only

The current eruption has destroyed all of the hillforest in the South Soufriegravere and Soufriegravere Hillswith the exception of a small (

c

200-ha) patch in thesoutheast of the South Soufriegravere Hills Severalnarrow fringes of degraded riparian forest in thelowlands of the north of the island continue to sup-port small populations of some forest bird speciesincluding Forest Thrush Other than this Montserratrsquosforest bird community is now confined to the CentreHills forest (

c

1400 ha) The Centre Hills wereaffected by major ashfalls in AugustndashOctober 1997There was a strong ashfall gradient within theCentre Hills such that sites in the south of the rangereceived ash to a depth of

c

50 mm whereas in thenorth of the range

c

5 mm of ash fell (MontserratVolcano Observatory unpubl data) Another sub-stantial ashfall occurred on 3 July 1998 A thirdmajor ashfall occurred on 29 July 2001 Accuratedata on ash levels are not available but ash depthswere similar to the 1997 events (CGR Bowden perscomm) On 1213 July 2003 a major dome collapsedeposited 150 mm of ash over southwestern parts ofthe Centre Hills decreasing to only 5 mm in thenortheast (Montserrat Volcano Observatory unpubldata) In the intervening periods there were numerousminor ashfalls In addition the large quantity of ashin the soil was frequently re-circulated by windaction during dry periods

During forest bird monitoring in the Centre Hills12 resident forest birds were recorded in sufficientnumbers to be statistically analysed (Table 1)Arendt

et al

(1999) report absolute populationdensities for these species in the Centre Hills based

300

B Dalsgaard

et al



et al


Bird census



et al





301



et al

(1999) and Hilton

et al





















302

B Dalsgaard

et al




deg

45

prime

N 62

deg

13

prime


r

= 093

P

lt 0001)

Data analysis


df


df



times



χ

2

statistic and hence

P




n


t


t

+1




lsquoCountrsquo

year

t

+1

=


+


year

t


year

t


year

t











303


a priori




lsquoCountrsquo

year

t

+1

=


+


times

lsquoAsh-levelrsquo

+


year

t


times




t



RESULTS

















Species

Recent ashfall















SpeciesEffect


Effect ( change)

Wald χ2 (df = 1) P






DISCUSSION



Species


Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

















Species



Canopy cover


























REFERENCES











































300

B Dalsgaard

et al



et al


Bird census



et al





301



et al

(1999) and Hilton

et al





















302

B Dalsgaard

et al




deg

45

prime

N 62

deg

13

prime


r

= 093

P

lt 0001)

Data analysis


df


df



times



χ

2

statistic and hence

P




n


t


t

+1




lsquoCountrsquo

year

t

+1

=


+


year

t


year

t


year

t











303


a priori




lsquoCountrsquo

year

t

+1

=


+


times

lsquoAsh-levelrsquo

+


year

t


times




t



RESULTS

















Species

Recent ashfall















SpeciesEffect


Effect ( change)

Wald χ2 (df = 1) P






DISCUSSION



Species


Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

















Species



Canopy cover


























REFERENCES













































301



et al

(1999) and Hilton

et al





















302

B Dalsgaard

et al




deg

45

prime

N 62

deg

13

prime


r

= 093

P

lt 0001)

Data analysis


df


df



times



χ

2

statistic and hence

P




n


t


t

+1




lsquoCountrsquo

year

t

+1

=


+


year

t


year

t


year

t











303


a priori




lsquoCountrsquo

year

t

+1

=


+


times

lsquoAsh-levelrsquo

+


year

t


times




t



RESULTS

















Species

Recent ashfall















SpeciesEffect


Effect ( change)

Wald χ2 (df = 1) P






DISCUSSION



Species


Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

















Species



Canopy cover


























REFERENCES











































302

B Dalsgaard

et al




deg

45

prime

N 62

deg

13

prime


r

= 093

P

lt 0001)

Data analysis


df


df



times



χ

2

statistic and hence

P




n


t


t

+1




lsquoCountrsquo

year

t

+1

=


+


year

t


year

t


year

t











303


a priori




lsquoCountrsquo

year

t

+1

=


+


times

lsquoAsh-levelrsquo

+


year

t


times




t



RESULTS

















Species

Recent ashfall















SpeciesEffect


Effect ( change)

Wald χ2 (df = 1) P






DISCUSSION



Species


Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

















Species



Canopy cover


























REFERENCES













































303


a priori




lsquoCountrsquo

year

t

+1

=


+


times

lsquoAsh-levelrsquo

+


year

t


times




t



RESULTS

















Species

Recent ashfall















SpeciesEffect


Effect ( change)

Wald χ2 (df = 1) P






DISCUSSION



Species


Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

















Species



Canopy cover


























REFERENCES























































Species

Recent ashfall















SpeciesEffect


Effect ( change)

Wald χ2 (df = 1) P






DISCUSSION



Species


Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

















Species



Canopy cover


























REFERENCES






















































SpeciesEffect


Effect ( change)

Wald χ2 (df = 1) P






DISCUSSION



Species


Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

















Species



Canopy cover


























REFERENCES














































DISCUSSION



Species


Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

Effect ( change)

Wald χ2 (df = 1) P

















Species



Canopy cover


























REFERENCES


















































Species



Canopy cover


























REFERENCES

































































REFERENCES











































Impacts of a volcanic eruption on the forest bird community of Montserrat, Lesser Antilles

Documents

Transcript of Impacts of a volcanic eruption on the forest bird community of Montserrat, Lesser Antilles