The Condor 96:368-380 0 The Cooper Omithological Society 1994

ANNUAL VARIATION IN AND PLANTS OF A TROPICAL SECOND-GROWTH WOODLAND ’

BETTE A. LOISELLE AND JOHN G. BLAKE Department of Biology, Universityof Missouri-St. Louis, 8001 Natural Bridge Road, St. Louis, MO 63121-4499

Abstract. We examined effectsof successionalchanges in vegetationon speciesfound in a young second-growthwoodland in Costa Rica over a four-year period (1985-1988). Significant changesin vegetation structure and plant composition occurred: numbers of small (2-5 cm dbh) trees, woody shrubs, and lianas increased, as did canopy cover; her- baceousshrubs declined in abundance.Total numbersand speciesof plantsproducing flowers pollinated by hummingbirds or fruits eaten by birds did not change,but significantchanges were noted in fruit production of individual plant species.Overall capture rates of birds in mist nets did not vary significantly among years but significant changeswere noted for individual species.Several bird speciescharacteristic of mature forest habitat first appeared in second-growthwoodland during latter years of the study; other species,more dependent on early successionalhabitats, decreasedin abundance or disappeared entirely from our study plot. Diets of some fruit-eating birds changed over time, apparently in responseto changesin abundance of different fruit species.This study suggeststhat changesin plant and bird populationsmay be relatively rapid in tropical second-growthhabitats. Suchresults have important implications for conservation;a rangeof successionalstages may be necessary to support the full complement of bird speciesthat can occur in a given area. Key words: Annual variation;Costa Rica; diet;frugivores;fruits;habitat selection;second- growth:succession: tropical birds.

INTRODUCTION dependent on the scale and severity of the dis- Disturbed or second-growthhabitats in tropical turbance. Anthropogenic disturbancesalso range regionsare important for many resident bird spe- over orders of magnitude in area, from small, cies, either on a permanent basis or for short- single-family plots to extensive cattle pasturesor term use of resources(Levey 1988a, 1988b; Blake monocrops. Short- and long-term effectson na- et al. 1990; Blake and Loiselle 199 1; see also tive plants and also vary. The ability of Gilbert 1980). Such habitats are important for disturbed habitats to regrow and attain some many migratory bird speciesas well (Martin 1985; measure of former complexity and diversity has Hutto 1989, 1992; Blake and Loiselle 1991, direct consequencesfor bird populations. 1992). Conversion of tropical foreststo pastures In caseswhere land use changeshave not re- or other land uses has increased the extent of sulted in severe environmental degradation, and second-growthhabitats in many regions of Cen- where seed sourcesare in relatively close prox- tral and South America. Despite the increasing imity, regrowth of vegetation in tropical habitats occurrenceof non-forested habitats in wet trop- can be extremely rapid following disturbance ical regions, few studieshave addressedhow suc- (pers. observ., Uhl and Jordan 1984, Uhl et al. cessional changesin habitat affect plant and an- 1990); related, concurrent changesin bird pop- imal communities (Uhl and Jordan 1984, Martin ulations are likely to occur in such habitats as and Karr 1986, Saldarriaga 1987, Hutto 1989, well. With the increasing occurrenceof disturbed Laurance 199 1). habitats, understanding the consequencesof such Natural disturbances range in size from single changesin habitat structure to bird populations treefall gaps to many square kilometers of hur- becomes increasingly important (Blankespoor ricane damage(e.g., papersin Walker et al. 199 1). 199 l), especially because some birds play a key Responsesof plants and animals to such distur- role in ecosystemfunction (e.g., as seed dispers- bances can be rapid, with the extent of change ers and plant pollinators; see Stiles 1985). Speciescomposition and abundance of birds and fruit-producing shrubsdiffer among second- ’ ’ Received 17 May 1993. Accepted 1 November growth habitats of different ages and between 1993. second-growthand forest habitats (Martin 1985;

13681 ANNUAL VARIATION IN TROPICAL BIRDS 369

Martin and Karr 1986; Levey 1988a, 1988b; (1958-1988, Organization for Tropical Studies, Hutto 1989; Lynch 1989, 1992; Blake and Loi- unpubl. data). The dry seasontypically lastsfrom selle 1991). Two basic approaches can be used January or February to April or May, with a to examine successionalchanges in plants and second,less pronounced dry seasonin September associatedanimals. One can either follow changes or October. December-April rainfall was below over time on a given study plot (e.g., Willis 1974) “normal” during three years of this study (i.e., or compare communities in a series of plots of 694 mm in 1985,743 in 1986,645 in 1987, and different known ages(e.g., Hutto 1989, Blankes- 1,129 in 1988; 30-year ave. rainfall for Decem- poor 1991). We use the first approach in this ber-April is 1,185 mm). study, examining changesin birds and plants over We established a study plot in January 1985 a four-year period (1985-1988) in a young (5- in an area of young second-growth vegetation 10 year old) anthropogenic second-growth hab- (approximately 40 ha in size). The study plot was itat in Costa Rica. We focus on species level on recent alluvial soil adjacent to Rio Sarapiqui responses(changes in activity levels as measured and was bordered by older (~20-30 year-old) by mist-net captures;changes in diet) to succes- second-growth. On approximately one-third to sional changesin habitat. We were particularly one-half of the study plot, Psidium guava (Myr- interested in determining what factors account taceae) was the dominant tree, creating an open for annual changesin bird communities and how canopy 5-6 m tall. Psidium was probably intro- quickly thesecommunities either begin to resem- duced to this area by seedsdefecated by horses ble mature forest or are attractive to forest spe- and/or cattle (G. S. Hartshom, pers. comm.). cies. To evaluate the influence of changesin re- Prior to abandonment in 198 1 (when the land sourceabundance on bird use of the habitat, we was purchasedby OTS, seeClark 1990) the area focus on fruit-eating birds and fruit-producing was used for pasture; some rapidly growing sec- plants. Many resident and migrant birds are at- ondary trees(e.g., Piper sp. [Piperaceae],Miconia tracted to the often abundant supplies of fruit sp. [Melastomataceae], Vismia sp. [Guttiferae]) that occur in second-growthhabitats (Martin and were already 3-4 m tall when the study began in Karr 1986, Blake and Loiselle 199 1). As fruit 1985. A few tall trees, such as Ficus sp. (Mora- abundance changes or as plant speciescompo- ceae; >20 m tall) and Nectandra membranacea sition changes, composition and abundance of (Lauraceae; >8 m tall), were left standing in the fruit-eating birds are also likely to change. plot. Early successionalplant and bird speciesfound in our study site resemble those found in similar METHODS habitat throughout the Atlantic forest region of Costa Rica (pers.observ.). Consequently, changes BIRDS in plant and bird communities observed at our We used mist nets (12-m, 4 shelf, 36-mm mesh) site likely occur in these other habitats as well. to sample birds. (See Blake and Loiselle [ 199 1] and Loiselle and Blake [ 199 l] for recent discus- STUDY AREA sions on use of mist nets.) We arranged 30 nets We conductedour researchat La Selva Biological in a grid that covered approximately 5 ha; Psid- Station (Organization for Tropical Studies [OTS]) ium trees were present on approximately one- in the lowlands of northeastern Costa Rica, near third of the study plot. Nets were centered on Puerto Viejo de Sarapiqui, Provincia Heredia points located 40 m apart on this grid (i.e., we (10”25’N, 84”Ol’W). La Selva encompassesap- did not select “good” sites); exact orientation of proximately 15 10 ha, of which about 64% is the nets varied among net-sites in order to min- primary (or slightly disturbed) forest(Clark 1990). imize disturbanceto vegetationwhen placing nets. The station also supports anthropogenic succes- We operated nets for four days (=l sample pe- sional habitats in various stagesof regrowth. To riod) every 5-6 weeks, from late December to the south, La Selva borders Parque National April (1985-1988) starting in January 1985. Braulio Carrillo (approximately 44,900 ha); con- There were three sample periodseach year. These tinuous forest exists from approximately 40 m sample periods corresponded to a time of year elevation at La Selva to >2,900 m in Braulio when latitudinal (primarily north temperate) and Carrillo (Pringle et al. 1984). altitudinal migrant birds were present at the site; Rainfall at La Selva averages 3,877 mm/year relatively few resident birds bred during this time 370 BETTE A. LOISELLE ANDJOHN G. BLAKE period (pets. observ.). Each net was operated on proximately the same time every year starting in only two alternate days each sample. Fifteen nets 1986. Four readings, one in each of the cardinal were opened per day, from sunrise to approxi- compass points, were taken at each end and at mately 13:00 hr, or as close to 7 hr/day as pos- the middle of each net lane (12 readings/net), sible. Captured birds were banded, weighed, and with the densiometer held at approximately 1.5 sexed and aged (if possible). We held all birds m; these 12 readings were then averaged to pro- (excepthummingbirds and raptors) for collection vide one canopy cover value for each net site per of fecal samples and then released birds at the year. point of capture. We determined diets of fruit- DATA ANALYSIS eating birds by examining fecal samples for fruit pulp and/or seeds(Loiselle and Blake 1990). We We used non-parametric Wilcoxon matched- identified >95% of seedsto plant speciesor ge- pairs test (Sokal and Rohlf 1981) to examine nus. Bird names follow American Ornithologists’ changesin habitat structure between 1986 and Union (1983, and supplements). 1989 by comparing vegetation characteristicsre- corded at all net-sites. For analyses of fruit and PLANTS flower abundance, we summed the number of Fruits and flowers were sampled during each of fruits and numbers of individual fruiting and the same periods that birds were netted. We sam- flowering plants over all 30 net sites during each pled fruits in two transects (2 x 12.5 m each) sample period (three sample periods/year) and located parallel to and 1 m away from the sides determined the number of specieswith flowers of each net (combined sample of 50 m*/net-site) or fruits. We used number of birds captured per (Blake et al. 1990). We identified and taggedall net-hour (one net open 1 hr = 1 net-hour) (x plants with flowers or fruit. Plant names follow 100) as an index of bird activity during each an unpublished checklistof La Selva plants (OTS, sample period. To compare capture rates over unpubl. data). We recorded presence of flowers time, we combined all captures from each net or flower buds and counted ripe and unripe fruits site for a particular sample to give a single value on each plant during each sample (following for that sample (three sample periods/year);mean methods outlined in Loiselle and Blake 1991). capture rate per year was then calculated from Flowers and flower buds were counted starting these three sample values. in 1986. Although fruits vary in size and nutrient We used one-way analysis of variance (ANO- content, we decided to use actual numbers of VA) to examine differences in fruit or flower fruits as our estimate of fruit available to birds. abundance or capture rate of birds among years. Most fruits used by understory birds in this hab- Data were tested for normality (Wilk-Shapiro itat are berries of the Melastomataceae and Ru- test) and homogeneity of variances(Bartlett ’s test) biaceae and, thus, offer mostly carbohydrate re- (Sokal and Rolhf 198 1) and were transformed wards and do not vary greatly in size (Loiselle (natural log) when necessaryprior to analysis. If and Blake 1990, 1991). We restricted our anal- transformed data did not meet assumptions of ysesto those plants known to be visited by birds parametric tests,a non-parametric test (Kruskal- for their fruit or nectar (basedon analysis of fecal Wallis) was used. samples [Loiselle and Blake 19901, personal ob- We determined if use of a particular fruit spe- servations of foraging birds, Croat 1978, Stiles cies by frugivorous birds varied among years. To 1980). do this, we summed the fecal sample occurrence Vegetation was sampled on two transects(3 x of seedsfrom that plant speciesby year and then 12.5 m each) located parallel to each side of each used a chi-square goodnessof fit test, with the mist net during 1986 and 1989. We measured null hypothesis being that frugivores fed on that and recorded the diameter at 1.5 m of all woody plant speciesequally among years (Loiselle and stems with a diameter 2 2 cm. Lianas 22 cm Blake 1990). Frugivorous birds have been shown thick were counted and measured, as were all in aviary trials to make consistentchoices among woody stems <2 cm dbh and 2 1.5 m tall, non- fruit species(e.g., Moermond and Denslow 1985). woody plants (L 1.5 m) (e.g., Reneufmia [Zin- In field situations, however, we expect that an- giberaceae];Heliconiu [Heliconiaceae]),and vines nual changesin fruit abundance and in compo- (~2 cm thick). Canopy cover was measured with sition of the frugivorous bird community likely a spherical densiometer (Lemmon 1956) at ap- result in annual differencesin visitation rates to ANNUAL VARIATION IN TROPICAL BIRDS 371 and fruit removal from a given plant species.To examine variation in relative use of different fruit speciesby individual bird speciesand by all fiu- givorous birds taken together, we summed the 25 occurrence in the diet of seeds of major plant species by year for a given bird species (or all 20 birds) and used chi-square (contingency table) teststo determine if that bird specieschanged its 15 t proportional use offruit speciesamong years. We \ combined diet data over two-year periods (1985 10 and 1986 vs. 1987 and 1988) for some species i-, becauseof small sample sizes. 5 RESULTS 0 PLANTS I I I Significant changes in vegetation structure and 1986 1987 1988 plant composition occurred over the course of FIGURE 1. Decreasesin canopyopenness (mean and this study. In general, vegetation at this site SE) detected at 1.5 m height in young second-growth changedfrom a nearly impenetrable massof dense from 1986 to 1988; means are based on canopy mea- vegetation at understory levels (i.e., ~2 m) to a suresat 30 net sites per year. relatively more open understory dominated by shrubs and small trees. This qualitative vegeta- plant species(species regularly visited by hum- tion change is demonstrated by the decreasein mingbirds) remained relatively constant among canopy openness(Wilcoxon Z = 4.77, P < 0.00 1; years (one-way ANOVA, F3,8= 0.33, P > 0.80). Fig. 1) and herbaceous(i.e., non-woody) shrubs Whereas, the mean number of flowering plants (Table 1). In contrast, there was significant re- declined markedly from 1985 to 1986, differ- cruitment of small (2-5 cm dbh) trees, woody encesamong years were not significant (one-way shrubs, and lianas (Table 1). Plant resourcesdi- ANOVA, FX,g= 1.56, P > 0.25) (Table 2). Sim- rectly used by birds (flowers, fruits) showed no ilarly, neither number of ripe fruits nor number significant changesoverall among years in over- of plants fruiting differed among years (F,,8 > all abundance. For example, number of flowering 0.20, P > 0.65) (Table 2).

TABLE 1. Mean number and standarderror (SE) of individuals per net-site (75 m2) for vegetation parameters measuredon transects(n = 30) in 1986 and on the same transectsin 1989. Resultsof Wilcoxon testscomparing means (by transects)between years are shown when significant.

1986 1989 Wilcoxon Vem3ation measure MCIII SE Meall SE test

Trees (dbh) 2-3 cm 4.2 0.65 11.3 1.17 P < 0.001 4-5 cm 2.1 0.41 3.3 0.47 P < 0.05 6-10 cm 4.9 0.64 3.9 0.44 11-15 cm 2.2 0.33 2.0 0.28 16-25 cm 0.7 0.20 0.8 0.17 26-50 cm 0.2 0.11 0.3 0.15 >50 cm 0.0 0.00 0.0 0.00 Lianas (diameter) 2-3 cm 0.5 0.20 1.9 0.47 P < 0.01 4-5 cm 0.1 0.07 0.4 0.18 P < 0.05 Woody shrubs and stems 61.2 3.73 85.6 7.19 P < 0.05 Vines and tendrils 16.0 3.18 23.5 3.83 Herbaceous shrubs 16.7 3.87 3.4 1.19 P < 0.01 Heliconia, Costus 2.0 1.73 3.4 1.49 372 BETTE A. LOISELLE ANDJOHN G. BLAKE

TABLE 2. Mean number of flowering plants or species(per 0.15 ha) and mean number of fruiting individuals or ripe fruits recorded in young second-growthwoodland based on three sample periods during each year. SE = standard error of samples during each year.

1985 1986 1987 1988 Mean (SE) Mean (SE) Mean (SE) Mean (SE)

Flowers Individuals 46.7 (10.97) 29.0 (7.5 1) 29.7 (6.89) 24.0 (5.03) Species 5.0 (0.0) 5.6 (0.88) 5.3 (0.33) 5.0 (0.58) Fruits1 Individuals 302(11.1) 337 (27.7) 412 (76.2) 327 (112.8) Ripe fruits 5,830 (1,773) 4,000 (928) 4,740 (2,442) 5,223 (3,281) Fruits2 Individuals 251 (20.6) 269 (27.3) 239 (19.9) 195 (11.9) Ripe fruits 4,767 (2,228) 2,949 (1,364) 690 (330) 1,219 (866)

’ ’ All species included. 2Psycharia pitfieri not included in totals; see text.

Changes at the plant species level, however, ined annual variation in capture rates for 19 spe- were more pronounced. During the four-year pe- cies that were relatively common on the plot and nod, 36 speciesof plants were recordedwith ripe that were active foragersat mist-net levels. These fruit from December-April. Of these 36, six spe- specieswere selectedbecause we felt that vari- cies accounted for over 75% of all ripe fruits and ation in capture rates among years reflected real plants with fruit. Abundance of three of those differencesin their activity on the plot. Capture six species(Conostegia subcrustulata, Clidemia rates of seven specieschanged significantly over dentata,Piper auritum) declined significantly (Fig. time (Table 3)-more speciesthan expected by 2). In contrast, number of fruiting Psychotriapit- chancealone (i.e., only one of 19 speciesexpected tieri plants (the most abundant fruit-producing to change due to chance alone). Capture rates of shrub on the plot) increased significantly from six speciesdeclined, whereasthat of Phaethornis 1985 to 1988 (Fig. 3). A similar increase in fruit- longuemareusincreased; these trends were con- ing was observed for Miconia afinis from 1985 sistent with knowledge of species’ preferred hab- to 1987; fruit production was lower, however, in itats. For example, all six species that showed 1988. Piper sancti-felicis showed no significant significant declines are speciescharacteristic of trend in fruit production over the four-year peri- young-second growth and edge in Costa Rican od on the study plot (Fig. 3). wet forestsand are rarely found in forest interior (Stiles and Skutch 1989, Blake et al. 1990). Mean BIRD ACTIVITY capturerates of severaladditional speciesshowed We captured 2,085 birds of 118 speciesduring declining or increasing trends over the four-year this study. Although fewer birds were captured period, but capture rates were often highly vari- over time, the decline was not significant (Table able within years (Table 3). Capture rates of eight 3). Capture rate of frugivores generally paralleled speciesshowed no annual change or no consis- capture rate of all birds among years. Captures tent trend. of nectarivores, on the other hand, declined Only three new species(i.e., speciesnot cap- markedly from 1985 to 1986 but remained fairly tured in any previous sample) were represented constant thereafter (Table 3). Changesin capture in mist-net capturesin 1988; 10 new specieswere rates of nectarivores tended to parallel changes represented in 1987, and 10 in 1986 (Table 4). in overall numbers of flowering plants in that In several instances, these “new” species (e.g., mean capture rate and mean number of flowering Legatus, Polioptila, Dendroica) were birds that plants both declined markedly from 1985 to 1986 occur primarily in forest canopy and that rarely and remained relatively stable thereafter. descend to mist-net levels. Several species, Overall, fewer bird specieswere captured in though, are birds found primarily in understory 1988 than in earlier years (Table 3). We exam- of undisturbed wet forests and their presencein ANNUAL VARIATION IN TROPICAL BIRDS 313

Conostegia subcrustulata 80

60 E 40 I m 20 ?I

Clidemia F 50 7 z 0 4o -0 6 30 3 v)

Piper \ :\ ...... _ __O____ auritum - 15 p A t u-l \ ..‘ I- 10 e‘ ....,‘ * F=28 ..,‘ : ..,‘ \ 0. ',(.. t 5

0 1985 1986 1987 1988 YEARS FIGURE 2. Mean number of individual fruiting plants or ripe fruits recorded during each year for three common plant species.We used total rather than ripe fruits for Piper auritum becausefruits do not always changecolor upon maturity and the height of plants made it impossible to test fruit ripenessby touch. Significant differencesin mean number of individuals or ripe fruits among years (ANOVA) are indicated. second-growthduring the latter yearsof this study annual changes in the abundance of that fruit might indicate the increasing suitability of this species.Specifically, consumption of Conostegia habitat for these birds (see Table 4). subcrustulataand Piper auritum fruits declined with time, whereas the reverse was true for Mi- DIETS OF FRUIT-EATING BIRDS conia afinis and Psychotria pittieri (Fig. 4). We examined annual variation in the diets of Moreover, fruit-eating birds as a whole (i.e., all fruit-eating birds, restricting our analyses to the frugivorous speciescombined) changedtheir rel- six common plants that accounted for over 75% ative intake of thesesix fruit speciesamong years of all fruits produced (see above). In general (4 year x 6 plant speciescontingency table, x2 changesin use of a given fruit speciesparalleled = 104.2, P < 0.0001). 374 BETTE A. LOISELLE AND JOHN G. BLAKE

28000

-----individuals ..... 0

8000

80 80 ’ 1 0 60 I_. 6o i m 40 40 5 0 20 20

0 1 0 1 985 1986 1987 1988 YEARS FIGURE 3. Number of fruits and individual fruiting plants for three common plant speciesin young second- growth.Significant differences in mean number of individualsor fruits amongyears (Kruskal-Wallis test) are indicated.We usedmaximum numberof fruitsrecorded during any oneyear rather than meannumber of fruits becauseof aseasonalfruit productionbetween samples within yearsto Psychotriapittieri and Miconia afinis.

Whether or not individual bird speciesshowed Miconia a&is fruits, and relatively fewer Co- significant differencesamong yearsin the relative nostegiasubcrustzdata fruits in their diets in later use of these six major fruit speciesappeared to years. Capture rates for both of thesebird species be related to the persistenceof thesebirds on the were relatively constant among years (Table 3), plot during the duration of this study. Two of suggestingthat these birds adjusted to changes the six most common fruit-eating birds, Muna- in fruit abundance by altering their diet. In con- cuscandei and Pipra mentalis, showedsignificant trast, Habia fuscicauda showed no variation in (x2 = 22, P < 0.01 for Manacus) or nearly sig- capture rates or relative use of fruit among years nificant (x2 = 9.3, P < 0.06 for Pipra) differences (Table 3, Fig. 5). Nonetheless, Habiu included among years in consumption of common fruits relatively fewer fruits of the two fruiting plants (Fig. 5). Both Pipru and, in particular, Munacus that were declining in abundance and included included relatively more Psychotria pittieri and relatively more of the two fruiting plant species

376 BETTE A. LOISELLE AND JOHN G. BLAKE

TABLE 4. Year that new specieswere first captured in young second-growthwoodland.

First capturedin 1986 First capturedin 1987 First capturedin I988

Geotrygon montana ’ Micrastur mirandollei ’ Dendrocincla fuliginosa ’ Phaenostictus mcleannani ’ Torgon rufus ’ Formicarius analis ’ Hylopezus fulviventri9 Xenops minutus Polioptila plumbeal,s Empidonaxjlaviventris2J Cymbilaimus lineatus Legatus 1eucophaiusJ’ Mionectes olivaceus1.4 Corapipo leucorrhoa+’ Aphanotriccus capitalis Thryothorus thoracicus Phaeothlypis fulvicauda Myadestes melanops,4’ Chlorophanes spizaS Dendroica magnolia Tachyphonus rufisS Tangara icterocephala4 Piranga olivacea3~s

’ Found primarily in primary forest. 2 Found primarily in older second-growth. ’ ’ Latitudinal migrant. a Altitudinal migrant. 5Primarily found in canopy.

that were increasing in abundance over the years vegetation in disturbed habitats regrew, second- (compare Figs. 2, 3, and 5). growth birds would have had to move in search Three speciesthat declined in abundance (cap- of new habitats. Even today, birds and other an- ture rates) over time (i.e., Ramphoceluspasseri- imals of second-growth habitats probably need nii, Saltator maximus, and Dumetella carolinen- to move in search of spatially and temporally sis; Table 3) did not differ in fruit use among variable disturbed habitats. As a consequenceof years (P > 0.30 in all cases).At least for Ram- these movements, and as results from our study phocelusand Saltator, fruits generally included indicate, second-growth bird and plant com- in their diet were from plant species that also munities can be quite dynamic over relatively declined in abundance over the same time period short time periods (Loiselle and Blake 1992). (Figs. 2, 5). These data suggestthat these bird We have observedtwo basic responsesof birds speciesmay be lessflexible in dietary preference, to changesin fruit resourcesin this second-growth and may have left the area due to changes in habitat, and in our other studies of fruit-fiugi- abundance of favored food resources(e.g., Piper vore systemsin Costa Rica. In conjunction with auritum, Conostegiasubcrustulata). Changes in seasonal declines in fruit resources, some obli- vegetation structure (Table 1) or resource gate frugivores move to new habitats or forest abundance, however, may also have influenced elevations where fruit resourcesare more abun- habitat selection by these birds and the impor- dant (Blake and Loiselle 199 1, Loiselle and Blake tance of these factors for this study is unknown. 199 1; also Wheelwright 1983). Such movements are temporary (few days to several months), with DISCUSSION specieslater returning to breeding habitats. That The Atlantic lowlands of Costa Rica have been birds often return to breeding habitats when fruit disturbed by humans for many hundreds of years abundance still is relatively low (F. G. Stiles and (Boucher et al. 1983). Charcoal found in the soils L. Rosselli, pers. comm.) indicates that factors of La Selva suggesta long history of human oc- other than fruit availability likely influence such cupation at this site as well (R. Sanford, pers. movements. A second response to changes in comm.). In addition, large-scale natural distur- fruit abundance is a concomitant switch in re- bances (e.g., hurricanes, earthquakes), likely re- sourceuse. This can either be a switch in relative sulted in relatively frequent, but spatially dis- use of resourcetypes (e.g., fruits to ) or a persed forest openings in the region (Yih et al. switch in relative use of fruit species(e.g., as in 1991). Prior to major human-induced changes Manacus, this study). Which strategy is em- in the landscape,many of the common bird spe- ployed by frugivores likely depends on the im- ciescaptured in young second-growthat La Selva portance of fruit in the diet (i.e., obligate vs. (e.g., Glaucis aenea, Ramphoceluspasserinii, Ar- opportunistic fruit-feeder), the seasonal avail- remonops conirostris, Saltator maximus) likely ability of fruit in the preferred (breeding) habitat were restricted to such natural disturbances. As (i.e., fruit production that is adequateyear-round ANNUAL VARIATION IN TROPICAL BIRDS 377

Conostegia subcrustulata Pip er auri turn

x2=15 6 P-CO.01

Miconia af finis Psuchotria pittieri

f2=22, P

85 86 07 88 85 86 a7 88 FIGURE 4. Percentoccurrence of four commonplant speciesin dietsof fruit-eatingbirds over four yearsin youngsecond-growth woodland. The numbersabove bars indicate the numberof fecalsamples where seeds of thisplant specieswere encountered. Results of Chi-squaretests that examineddifferences among years in number of fecalsamples containing these fruits are shown(see Methods). for maintenance or that drops below some mature forest, and most individuals captured threshold), and the presenceor absence of spe- during this study were transients from lowland cialized breeding systems(see below). mature or foothill forest. By contrast, active Bird speciestypically found only in early sec- Manacus candei and Phaethornis superciliosus ond-growth habitats (e.g., Ramphoceluspasseri- leks were located near our second-growth plot, nii) may have left the site both as a consequence in similar habitat. Such leks are traditional and of resourcechanges (declines in fruit abundance) persist over many years (Snow 1962, Stiles and and as a result of habitat changes(denser vege- Wolf 1979, pers. observ.). Consequently, species tation, less open canopy). In contrast, species with specialized breeding systems may be more typically found in second-growthhabitats of var- likely to remain in second-growth habitats over ious ages(e.g., Manacus), may be less sensitive time, despite changesin vegetation structure and to short-term habitat changesand more likely to resource abundance. It should be remembered, persist in such areas over longer periods. however, that the abundance of Heliconia and Four of the eight common bird species that Costus, two important plant genera for Phae- showedlittle changein activity among years have thornishummingbirds, did not changeduring this lek breeding systems.Mionectes oleagineusand study (Table 1). Pipra mentalis, however, are generally birds of This second-growth woodland was approxi- 378 BETTE A. LOISELLE AND JOHN G. BLARE

Manacus candei Pipra mentalis t 1

25 -

amphocelus passerin Saltator maximus 95 49 29

25

0 Dumetella carolinensis Habia fuscicauda

100

75

50

25

I 0 l- I 85 86 87 88 85 86 87 88

BConostegia NPiper sancti-felicis Clidemia

mPsychotria t_i Miconia m Piper auritum

FIGURE 5. Percent occurrenceof fruiting plants in the diets of common fruit-eating birds over a four year period in young second-growthwoodland. Only common plants that regularly occurred in the diets of these birds are shown in the figure. For Habia fuscicaudaand Dumetella carolinensis,data were combined for two- year periods (1985 and 1986 combined and shown in 1986 column; 1987 and 1988 combined and shown in 1987 column) to alleviate sample size problems. ANNUAL VARIATION IN TROPICAL BIRDS 319 mately 10 years old by the end of this study, and cycle), providing a spatial mosaic of patches we saw evidence that birds typical of mature within the forest that is relatively constant over forest understory were colonizing the habitat time. (Table 4). For example, we believe that some species(e.g., Trogon rufis, Dendrocinclafuligi- ACKNOWLEDGMENTS nosa, Formicarius analis, and Xenops minutus), We thank J. Hunt, R. Hutto, R. Marquis, and E. Wie- more commonly found in older forest, colonized ner for many helpful discussionsand comments on this the plot during this study. We do not know for manuscript:We greatlyappreciate the help and support certain, however, if these species were able to of David and Deborah Clark and Ronald Suarez for facilitating our researchat EstacionBiolbgica La Selva successfullybreed in this habitat. At least Trogon and the numerousresearchers and staffwho have made rufus and Formicarius analis appeared to set up work at this tropical site so enjoyable over the years. territories as regular singing behavior was re- We thank the Organization for Tropical Studies, the corded during censusesin later years. Nonethe- DepartmentofZoology-University of Wisconsin(Guyer less, the rapid changein the vegetation, together Fellowship), the National Geographic Society,and the Douroucouli Foundation for their financial support. with the relatively early colonization by some forest understory speciessuggests that rapidly re- LITERATURE CITED generating second-growth is an important alter- native habitat for some forest species. Preser- AMERICANORNITHOLOGISTS UNION. 1983. Check- vation of some speciesmay be enhanced by the list of North American birds. American Omi- tholoaists’ Union, Washington. DC. presenceof such habitats. It is important to rec- BLAKE,J.-G., AND B. ’ A. LOIS~LLE.' 1991. Variation ognize, however, that the majority of forest spe- in resourceabundance affects capture rates ofbirds cies have not been found on this plot, indicating in three lowland habitats in Costa Rica. Auk 108: that second-growthhabitats, while beneficial for 114-130. BLAKE,J. G., AND B. A. LOISELLE.1992. Habitat use some species,will not substitute for mature for- bv Neotrouical migrants at La Selva Biological est. Siation and BraulioCarrillo National Park, Costa Naturally disturbed and human-disturbed Rica, p. 257-272. In J. M. Hagan, III and D. W. habitats are prevalent in this region of Central Johnston[eds.], Ecologyand conservationofNeo- America and, especially in regenerating forests, tropical migrant landbirds. Smithsonian Institu- tion Press, Washington, DC. support many plant and speciesthat do BLAKE,J. G., B. A. LOISELLE,T. C. MOERMOND,D. J. not occur inside mature forest. Thus, they con- LEVEY,AND J. S. DENSLOW. 1990. Quantifying tribute to the region’s biodiversity (Brown and abundanceof fruits for birds in tropical habitats. Brown 1992). Moreover, such regenerating hab- Stud. Avian Biol. 13:73-79. BLAKE,J. G., F. G. Srnns, ANDB. A. LOISELLE.1990. itats often provide important resources (fruit, Birds of La Selva: habitat use, trophic composi- nectar) for birds (and other animals) during pe- tion, and migrants,p. 16l-l 82. In A. Gentry [ed.], riods of food shortage in more mature forests Four Neotropical rainforests. Yale Univ. Press, (Stiles 1983, Terborgh 1983, Levey 1988a, Blake New Haven, CT. et al. 1990, Blake and Loiselle 199 1; also Gilbert BLANKESPOOR,G. W. 1991. Slash-and-bum shifting agricultureand bird communities in Liberia, West 1980). Maintenance of a mosaic of habitat types Africa. Biol. Conserv. 57:4 l-7 1. in a particular region or reserve may require sub- BOUCHER,D. H., M. HANSEN, S. RISCH, AND J. H. stantial and active management intervention. As VANDERMEER.1983. Agriculture: introduction, protected forests become more isolated and re- p. 66-72. In D. H. Janzen [ed.], Costa Rican nat- ural history. Univ. Chicago Press, Chicago, IL. stricted in , and to the extent BROWN,K. S., JR., AND G. G. BROWN. 1992. Habitat that surrounding land use becomes more intense alteration and speciesloss in Brazilian forests, p. and resemblesnatural ecosystemsto a lesserde- 119-142. In T. C. Whitmore and J. A. Sayer[eds.], gree, the management of natural forestsfor bio- Tropical deforestation and species extinction. diversity may entail the creation of some dis- Chapman and Hall, London, England. CLARK, D. B. 1990. La Selva Biological Station: a turbed areas (cf. Gilbert 1980). Dependent on blueprint for stimulating tropical research, p. 9- the size of the protected forest patch, disturbed 27. In A. Gentry [ed.], Four Neotropical rainfor- areas can be allowed to mature naturally while ests. Yale Univ. Press, New Haven, CT. other areas are disturbed, thus providing a tem- CROATT. B. 1978. Flora of Barro Colorado Island. Smithsonian Institution Press, Washington, DC. poral and spatial mosaic of habitat patches. Al- GILBERT,L. E. 1980. Food web organizationand con- ternatively, managed areas may need to be dis- servation of Neotropical diversity, p. 1l-34. In turbed periodically (perhaps on a 15-20 year M. E. SoulCand B. A. Wilcox [eds.],Conservation 380 BETTE A. LOISELLE AND JOHN G. BLAKE

biology: an evolutionary-ecological perspective. HARTSHORN,G. SCHATZ,F. G. STILES,C. GOMEZ, Sinauer Associates,Sunderland, MA. AND M. RODRIGUEZ. 1984. Natural history ob- Hurro, R. L. 1989. The effect of habitat alteration servationsand ecologicalevaluation of the La Sel- on migratory land birds in a West Mexican tropical va protection zone. Brenesia 22: 189-206. deciduousforest: a conservationperspective. Con- SALDARRIAGA,J. G. 1987. Recovery following shift- serv. Biol. 3:138-148. ing cultivation, p. 24-33. In C. F. Jordan [ed.], HUTTO, R. L. 1992. Habitat distributions of migra- Amazonian rain forests:ecosystem disturbance and tory landbird speciesin western Mexico, p. 22 l- recovery. Springer-Verlag,New York. 239. In J. M. Hasan. III and D. W. Johnstonleds.1. SNOW.D. W. 1962. A field studv of the Black and Ecology and conservation of Neotropical migrant White Manakin, Manacus rna~acus, in Trinidad. landbirds. Smithsonian Institution Press, Wash- Zoologica 47:65-104. ington, DC. SOKAL,R. R., AND F. J. ROHLF. 1981. Biometry. LAURANCE,W. A. 1991. Ecologicalcorrelates of ex- Second edition. W. H. Freeman, San Francisco. tinction pronenessin Australian tropical rain for- STILES,F. G. 1980. The annual cycle in a tropical wet est mammals. Conserv. Biol. 579-89. forest hummingbird community. Ibis 122:322- LEMMON,P. 1956. A spherical densiometer for esti- 343. mating forest overstory density. Forest Sci. 2:3 14- STILES,F. G. 1983. Birds: introduction, p. 502-530. 319. In D. H. Janzen [ed.], Costa Rican natural history. LEVEY,D. J. 1988a. Spatial and temporal variation Univ. of Chicano Press. Chicago, IL. in Costa Rica fruit and fruit-eating bird abun- S~nn.s,F. G. 1985: On the role ofbirds in the dy- dance. Ecol. Monogr. 58:25l-269. namics of Neotropical forests, p. 49-59. In A. W. LEVEY,D. J. 1988b. Tropical wet forest treefall gaps Diamond and T. Lovejoy [eds], Conservation of and distributions of understory birds and plants. tropical forest birds. International Council Bird Ecology 69:1076-1089. Preservation, Cambridge, England. LOISELLE,B. A., AND J. G. BLAKE. 1990. Diets of Sn~n.s,F. G., AND A. F. SKUTCH. 1989. A guide to understory fruit-eating birds in Costa Rica. Stud. the birds of Costa Rica. Cornell Univ. Press, Ith- Avian Biol. 13:91-103. aca, NY. LC~SELLE,B. A., AND J. G. BLAKE. 1991. Resource STILES,F. G., AND L. L. WOLF. 1979. Ecology and abundanceand temporal variation in fruit-eating evolution of lek mating behavior in the Long-tailed birds along a wet forest elevational gradient in Hermit Hummingbird. Omithol. Monogr. No. 27. Costa Rica. Ecoloav 72: 180-193. TERBORGH.J. 1983. Five New World mimates: a L~ISELLE,B. A., ANDJTG. BLAKE. 1992. Population study ‘in comparative ecology. Princeton Univ. variation in a tropical bird community: implica- Press, Princeton, NJ. -_ tions for conservation. Bioscience 42:838-845. UHL. C.. AND C. F. JORDAN. 1984. Successionand LYNCH,J. F. 1989. Distribution of overwintering Ne- nutrient dynamics following forest cutting and arctic migrants in the Yucatan Peninsula, I: gen- burning in Amazonia. Ecology 65:1476-1490. eral patterns of occurrence.Condor 9 1:515-544. UHL, C., D. NEPSTAD,R. BUSCHBACHER,K. CLARK,B. LYNCH,J. F. 1992. Distribution of overwintering Ne- KAUFFMAN,AND S. SUBLER.1990. Studiesofeco- arctic migrants in the Yucatan Peninsula, II: use systemresponse to natural and anthropogenicdis- of native and human-modified vegetation,p. 178- turbances provide guidelines for designing sus- 195. In J. M. Hagan, III and D. W. Johnston[eds.], tainable land-use systemsin Amazonia, p. 24-42. Ecologyand conservationof Neotropical migrant In A. B. Anderson [ed.], Alternatives to defores- landbirds. Smithsonian Institution Press, Wash- tation: stepstoward sustainableuse of the Amazon ington, DC. rain forest. Columbia Univ. Press,New York, NY. MARTR‘I, T. E. 1985. Selection of second-growth WALKER,L. R., N.V.L. BROKAW,D. J. LODGE,AND R. woodlands by frugivorousmigrating birds in Pan- B. WAIDE [EDS.]. 1991. Ecosystem, plant, and ama: an effect of fruit size and density?J. Trop. animal responsesto hurricanesin the . Ecol. 1:157-170. Biotropica Special Issue 23(4A). MARTIN, T. E., AND J. R. I(ARR. 1986. Temporal WHEELWRIGHT,N. T. 1983. Fruits and the ecology dynamics of Neotropical birds with special ref- of the Resplendent Quetzal. Auk 100:286-301. erenceto frugivoresin second-growthwoods. Wil- WILLIS,E. 0. 1974. Populationsand local extinctions son Bull. 98:38-60. of birds on Barro Colorado Island, Panama. Ecol. MOERMOND,T. C., AND J. S. DENSLOW.1985. Neo- Monogr. 44:153-169. tropical frugivores: patterns of behavior, mor- YIH, K., D. H. BOUCHER,J. H. VANDERMEER,AND N. phology, and nutrition with consequencesfor fruit ZAMORA. 1991. Recovery of the rain forest of selection. Omithol. Monogr. 36:865-897. southeasternNicaragua after destruction by hur- PRINGLE,C., I. CHACON,M. GRAYUM,H. GREENE,G. ricane Joan. Biotropica 23: 106-l 13.