Distribution and Relative Abundance Patterns of Columbids in Puerto Rico ’

Home , Geotrygon, Zenaida dove, Zenaida doves

The Condor 94~224-238 0 The CooperOrnithological Society 1992

DISTRIBUTION AND RELATIVE ABUNDANCE PATTERNS OF COLUMBIDS IN PUERTO RICO ’

FRANK F. RIVERA-MILAN Department of Natural Resources,Scientific Research Area, TerrestrialEcology Section, P.O. Box 5887, Pta. de Tierra, PR 00906

Abstract. From July 1986 to January 1989 standardized call and sight counts were conducted to examine the distribution and relative abundance patterns of Plain Pigeons (Columba inornata), White-crowned Pigeons (C. leucocephala),Scaly-naped Pigeons (C. squamosa),Zenaida Doves (Zenaida aurita), White-winged Doves (Z. nsiatica), Mourning Doves (Z. macroura),Common Ground-Doves (Columbinapasserina),Ruddy Quail-Doves (Geotrygonmontana), Key West Quail-Doves (G. chrysia), and Bridled Quail-Doves (G. rnystacea)during the different seasonsof the year in the dry, moist and wet zones of Puerto Rico. The relative abundance of all columbids peaked during the spring-summer period (March-August) in the life zones. The relative abundanceof Zenaida Doves and Common Ground-Doves peaked during the spring-summer, but a second noticeable peak occurred during the fall (September-November) in the dry zone. The seasonalpatterns of relative abundancewere detectable at local as well as at regional scalesof sampling, regardlessof the observer involved and the counting technique applied. The detected patterns are not consideredto be either the result of sampling error or of statistical artifacts. The seasonal patterns of relative abundance in the three major life zones probably represent the most general and repeatableaspect of the population dynamics of columbids in the island. Two components of the environment of columbids are considered of principal importance in determining their seasonalrelative abundancein the life zones: (1) rainfall in the dry zone, and (2) fruiting phenology of key tree speciesin the moist and wet zones. Kev words:Columbidae: distribution: relative abundance; life zones:seasons; standardized counts;Puerto Rico.

INTRODUCTION ing are probably the two principal causesof en- Of the 284 speciesof pigeonsand doves (Colum- dangerment of columbid populations in Puerto bidae) recognized world-wide (Goodwin 1983) Rico (Perez-Rivera and Collazo-Algarin 1976, 10 speciesare native to Puerto Rico: Plain Pigeon Wiley and Wiley 1979, Raffaele 1989, Rivera- (Columba inornata), White-crowned Pigeon (C. Milan 1990). Long-term studies of the popu- leucocephala), Scaly-naped Pigeon (C. squamo- lation dynamics of columbids are needed to sa), Zenaida Dove (Zenaidu aurita), White- generate recommendations for an integrated winged Dove (Z. asiatica), Mourning Dove (Z. management plan (Rivera-Milan et al. 1990). macroura), Common Ground-Dove (Columbina In July 1985 a sampling scheme was estab- passerina), Ruddy Quail-Dove (Geotrygon mon- lished to determine the distribution and relative tuna), Key West Quail-Dove (G. chrysia), and abundance of Zenaida Doves and Scaly-naped Bridled Quail-Dove (G. mystacea). The Rock Pigeonsin the three major life zones ofthe island Dove (C. livia) and the Ringed Turtle Dove (Collazo-Algarin et al., 1985). Five principal (Streptopelia risoria) are non-native speciesfound sampling units (PSUs) 8 km long each were ran- mainly in domestic and semi-domestic states domly selected per life zone to conduct stan- (Raffaele 1989; Rivera-Milan, pers. observ.), and dardized roadside counts; i.e., a total of 15 PSUs were not considered in this investigation. were established following the rationale of a Four species are legally hunted during fall stratified random design (see Cochran 1977). In (September-November): Scaly-naped Pigeons, July 1986 the sampling schemewas expanded to Zenaida Doves, White-winged Doves, and include both game and non-game columbid spe- Mourning Doves. Habitat loss and illegal hunt- cies, and detection distanceswere defined as unfixed and fixed at 60 m for call and sight counts (Rivera-Milan 1990). I Received 18 July 1991. Accepted 13 September Here I provide data from unfixed counts and 1991. discuss the resulting distribution and relative

12241 COLUMBIDS IN PUERTO RICO 225 abundance patterns of columbids at different variable sizes of successionaltree species with spatial (PSUs, life zones, islandwide) and tem- rapid growth rates are common in disturbed ar- poral (seasons,years) scalesof resolution. I had eas. two specific objectives: (1) to examine the dis- The subtropical wet zone encompassesabout tribution and relative abundance patterns of co- 24% (2,124.g km2) of the total land area of the lumbids during the different seasonsof the year island. Mean annual precipitation ranges from in the three major life zones of the island, and 2,200 to 4,000 mm. Rainfall usually peaks dur- (2) to determine how dependent are the patterns ing May, and from August to November. Epi- on the spatiotemporal scaleunder consideration. phytes are widespread,and usually provide com- Special attention was given to the most general plete cover to branches. Successionaltree species and repeatable patterns of aural and visual de- have rapid growth rates, forming closedcanopies tections. The goal of this investigation is to detect at about 20 m. Most plant speciesare evergreen real population patterns and proposetestable ex- and sclerophyllous. In karstic areas, the vegeta- planations of the controlling processesinvolved. tion is more xeric than expected due to water percolation and runoff of soils.

STUDY AREAS AND METHODS STANDARDIZED COUNTS LIFE ZONES Five PSUs 8 km long each were established per Puerto Rico, the smallest (8,80 1.9 km*) and east- life zone along secondaryand tertiary roads that ernmost of the Greater Antilles (17”55’-18”35’ were lightly traveled during the early morning north, 65”37’-67”17’ west; Fig. l), lies within the hours (Table 1). Six sampling stations, spacedat subtropical latitudinal region, and contains six regular intervals of 1.6 km, were established per life zones (Ewe1and Whitmore 1973). The three PSU. Counts were conducted for 3 min at each major life zones have been classifiedas subtrop- sampling station. During each count, the number ical dry, subtropical moist, and subtropical wet, of pigeons and doves heard calling, as well as the and comprise approximately 98% of the total number seen(e.g., singles,pairs, or flocks ofvari- land area. Presently, the forest area of the island able sizes),were recorded,regardless of detection encompasses about 2,790 km2 (Birdsey and distance (= unfixed). Each observation was clas- Weaver 1982). sified as aural or visual depending on the initial The subtropical dry zone encompasses ap- form of detection. A separaterecord was kept of proximately 14% (1,2 16.4 km2) of the total land the individuals seen per species while driving area of the island. Mean annual precipitation between stations. Aural detections while driving rangesfrom 600 to 1,100 mm. Rainfall normally were considered negligible, becauseof their low peaks during May and from September to No- frequency of occurrence. Driving speedbetween vember. The vegetation is characteristically de- stations normally fluctuated between 16-32 kph ciduous or semi-deciduous on most types of soil. (1 O-20 mph) dependingon problems suchas road Plant speciescommonly have thorns and spines. conditions. Epiphytes (e.g., bromeliads and vines) are com- Counts were conducted from 06:OO to lo:30 mon, but they rarely provide much cover over hours. Surveys were suspendedwhen it was rain- branches. Tree heights rarely exceed 15 m, and ing or the wind speed exceeded 12 mph (19.2 their crowns are usually broad and flattened with kph). Standard data sheetsofunfixed counts were sparsefoliage. similar to those used by the U.S. Fish and Wild- The subtropical moist zone encompassesabout life Service (USFWS) for Mourning Doves (Dol- 60% (5,326.l km2) of the island. Mean annual ton 1985). precipitation ranges from 1,100 to 2,200 mm. The PSUs were sampledat leastonce per month Rainfall usually peaks in May and from August by two experienced observers.The sampling pe- to November. Tree heights commonly reach up riod covered from July 1985-June 1988 in the to 20 m, and their crowns are usually rounded moist and wet zones; and from July 1985-Jan- in shape. Many plant speciesare deciduous dur- uary 1989 in the dry zone. Counts were not con- ing the dry season(usually from January to April). ducted in urbanized areas,although some species Epiphytes are common and provide important are well adapted to urban habitats (e.g., Common cover over branches. Grasses characteristically Ground-Dove and Zenaida Dove). cover the ground after deforestation. Patches of Some distant calls of quail-doves could have 226 FRANK F. RIVERA-MILAN

I 650 6b” WEST

ATLANTIC OCEAN

t N

PUERTO RICO * I .i?’ ’ VIRGIN.? o ISLAND-i % o ‘0

150 NORTH 0 CARIBBEAN SEA Q 0 0 b -0 ’ TOBAGO TRINIDAD 8

VENEZU E L A

D BD 160 COLOMBIA

1 BOOO’

FIGURE 1. Caribbeanregion showing the locationof PuertoRico with respectto the restof the Greaterand LesserAntilles, and the principalsampling units (PSUs)under study. been confuseddue to the presenceof background speciesbefore conducting statistical analyses of noise (e.g., near rivers). Moreover, Bridled and counts (see below). Key West Quail-Doves could have been con- fused in some instances, because of their body RA1NFALL size and plumage characteristics (see Raffaele Rainfall data for the dry zone were obtained from 1989). The problem was circumvented by pool- pluviometers in the xerophytic forest of Gulnica ing aural and visual detections of the quail-dove (PSU 14; Canals, unpubl. data) and the dry coast- COLUMBIDS IN PUERTO RICO 221

TABLE 1. Principal sampling units (PSUs) used to estimate the seasonalrelative abundanceof columbids in the three major life zones of Puerto Rico.

Location No. PSU Habitat type

Subtropical moist zone (1) Aibonito: roads 7718, 14, 723, & 558 PSU 3 Moist montane forest (2) Guayama: road 708 PSU 25 Moist montane forest (3) Barceloneta:tertiary roads PSU 32 Wetland, agricultural, & moist limestone forest (4) Mayagiiez: road 108 PSU 34 Moist montane forest (5) Isabela: road 112 PSU 35 Moist limestone forest Subtropical wet zone (1) Rio Grande: roads 191 & 9966 PSU 2b Upper & lower montane rain forest (2) Barranquitas:road 143 PSU 5 Wet montane forest (3) Ciales: road 143 PSU 6 Wet montane forest (4) Cayey: tertiary road and road 184 PSU lob Wet montane forest (5) Cayey: road 742 PSU 12 Wet montane forest (6) Adjuntas: road 128 PSU 21c Wet montane forest (7) Maricao roads 120, 365 & 366 PSU 23c Wet montane forest Subtropical dry zone (1) Lajas: tertiary road PSU lc Agricultural & dry coastal forest (2) Guanica: tertiary road PSU 14 Xerophytic forest (3) Cabo Rojo: tertiary road & road 302 PSU 15 Dry coastal forest (4) Guayama: road 706 PSU 27” Agricultural, xerophytic, & dry coastal forest (5) Cabo Rojo: tertiary road & road 303 PSU 28 Dry coastal forest (6) Lajas: road 324 PSU 29 Dry coastal forest (7) Santa Isabel: tertiary road & road 545 PSU 36b Agricultural & dry coastal forest

- Habitat typesdescribed according to rainfall and generalcomposition of the vegetation(see Ewe1 and Whitmore 1973). bSampling in theseF ’SUs was initiated during February 1987. rThese PSUs were not sampledafter January1986. Substituteswere selected at random from a pool of 36 possiblechoices (see b above). al forest of Cabo Rojo (PSU 15; USFWS, Carib- bean Islands National Wildlife Refuge, unpubl. FRUITING PHENOLOGY data). Rainfall patterns were examined as totals The data gathered by Cardona et al. (1986) were and averages per month, based on five and eight used to depict the fruiting phenology patterns of years of data available from PSUs 14 and 15, key tree speciesin the subtropical wet forest of respectively. Call counts of Zenaida Doves in the Rio Abajo. The following tree specieswere con- dry zone covered a period of 43 consecutive sidered important food sourcesfor pigeons and months. Rainfall data were divided into six doves in Rio Abajo (seeCardona et al. 1986, and month intervals to study the possibility of time literature therein): Apple Rose (Syzygium jum- lags in the responsiveness of calling Zenaida bos), Puerto Rican Royal Palm (Roystoneu bor- Doves during the nesting season (Rivera-Milan inquena), India Laurel Fig (Ficus citr$Aia), 1990). Time lags were studied up to two years Trumpet Tree (Cecropia schreberiana), Night previous to the detected calling peaks of doves Shade (Solanum torvum), Martinique Prickly- in PSUs 14 and 15. Rainfall was also studied on Ash (Zanthoxylum martinicense),Mountain Im- a seasonal basis, using years as replicates (n = mortelle (Erythrina poeppigiuna), Rabo de Ra- 3). Following Faaborg et al. (1984), DeSante and ton (Gonzalagunia spicata), and False-Mastic Geupel(l987), and Faaborg and Arendt (in press), (Mastichodendronfoetidissimum). I studied the departures from normal (DFN) were calculated fruiting phenology pattern of S. jumbos from July as absolute values resulting from the subtraction 1987-June 1988 in the PSUs of the wet and moist of ‘actual’ from ‘normal’ rainfall. DFN values zones and Carite Lake riparian forest (Cayey). were used to detect negative or positive responses Binoculars were used to make observations from of calling Zenaida Doves to rainfall extremes. the ground in 8 0.1 ha strip-transects located in 228 FRANK F. RIVERA-MILAN the riparian vegetation bordering Carite Lake, fruiting phenology data gathered by Cardona et and 60 circular plots (radius = 60 m) located in al. (1986) at Rio Abajo forest. Eight categories the PSUs of the moist and wet zones. I also con- (k - 1) were used to define the availability of ducted phenological observations of Z. martin- immature and mature fruits (pooled) of key tree icense,C. schreberiana,D. morototoni, R. borin- species on a seasonal basis at Rio Abajo (see quena, F. citrifolia, Erythrina poeppigiana, Kleinbaum and Kupper 1978). The availability Camasey (Miconia prasina) and Myrcia spp. from of immature and mature fruits of S. jumbos in July 1987-June 1988 in 7 0.1 ha strip-transects the PSUs of the wet and moist zones and Carite in patches of secondary growth forest in Cidra Lake riparian forest was categorizedas: 1 = low and Comerio. (O-25%), 2 = moderate (25-50%), and 3 = high (> 50%); similar categorieswere used for the phe- STATISTICAL ANALYSES nological observations made in Cidra and Come- Seasonal statistics were computed by pooling rio. Linear and multiple regressionswere used counts per month per life zone (n = 5 PSUs) as in evaluating the relationships between depen- follows: winter: December, January and Febru- dent (e.g., calling birds) and independent (e.g., ary; spring:March, April and May; summer: June, rainfall and fruit availability) variables. July and August; and, fall: September, October and November. RESULTS Statistical analyses of counts need to account Data for Plain Pigeons, White-crowned Pigeons, for the repeated measurements made per PSU and Bridled Quail-Doves are discussedonly in (P. H. Geissler, K. H. Pollock, pers. comm.). The relation to the most general and repeatable pat- problem is that the same sampling units (PSUs) terns of aural and visual detections of all colum- were visited month after month. Hence, positive bids in the life zones. In mainland Puerto Rico, correlations among visits were expected(see Wi- these species are uncommon outside a few lo- ner 197 1; Gurevitch and Chester 1986; Beal and calities. For example, Plain Pigeons outside the Khamis 1990, 199 1; Quinn and Keough 199 1). municipality of Cidra (in which density estimates The counts can be expectedto be more correlated were in the order of 0.4-l .2 birds per kmz) were (dependent) as visit intervals decrease (Gure- detectedonly twice after 2,400 point counts con- vitch and Chester 1986). These positive corre- ducted in 90 sampling stations (Rivera-Milan lations can be the cause of a serious underesti- 1990). Bridled Quail-Doves were infrequently mation of the variance associatedwith the counts, detected in PSUs 14 and 32, whereas White- and of an increase in the Type I error rate (alpha) crowned Pigeons were infrequently detected in associatedwith the statistical analyses conduct- PSUs 2, 15, 32, 34, and 35. ed. Since I was interested in examining counts ZENAIDA DOVES conducted at intervals of one month, analysesof A significant calling peak was detected for the variance (ANOVAs) with repeated measures at spring months after 3 years of data collection in one factor (e.g., seasons)were conducted. Upon the life zones (F = 12.293; df = 3, 108; P < significance, paired t-tests and Student-New- 0.0001). In the wet and moist zones calling ac- man-Keuls (SNK) multiple comparisons proce- tivity extended from January until at least Au- dures were performed as follow-up tests (Sokal gust, whereas calling activity from September to and Rohlf 198 1, Day and Quinn 1989). Signifi- November was negligible. In the dry zone calling cance was accepted at P 5 0.05. activity peaked between April and June, but a After finding several significant F-max and second noticeable peak occurred between Sep- Bartlett’s tests, and significant positive correla- tember and November (F = 2.482; df = 3, 36; tions between means and standard deviations of, P = 0.076). Visual detections in the life zones for example, calling Zenaida Doves (r = 0.93 1, peakedduring the spring-summer period (March- df = 25, P < 0.001) and Scaly-naped Pigeons (r August: F = 11.441; df = 3, 9; P = 0.002). = 0.883, df = 25, P < 0.001) counts were log- The number of calling birds detected in the transformed (log,,x + 0.5). Counts based on au- dry and moist zones was significantly different ral and visual detections appeared to be reason- from the wet zone (F = 9.522; df = 2, 36; P = ably close to a log-normal distribution. 0.0005). When the wet zone was excluded from Dummy variables were generated from the the analyses,the difference in the number of call- COLUMBIDS IN PUERTO RICO 229 ing birds detected in the dry and moist zones 12.275; df = 1, 49; P = 0.00 l), and summer (F became non-significant (F = 0.035; df = 1, 24; = 4.073; df = 1, 49; P = 0.049). P = 0.854). Similarly, the number of visual detections in the dry and moist zones were signif- COMMON GROUND-DOVES icantly different from the wet zone (F = 157.226; Two years of data collection revealed a signifi- df = 2, 3; P = 0.0009). However, the difference cant calling peak during spring (F = 9.564; df = in the number of visual detections between the 3, 72; P < 0.0001). Calling activity peaked be- dry and moist zones was non-significant (SNK; tween April and June; but, as in the caseofzenai- P > 0.05). The seasonal patterns of aural and da Doves, a secondnoticeable peak occurred be- visual detections of Zenaida Doves in the life tween September and November in the dry zone zones remained unchanged during the sampling (F = 2.550; df = 3, 18; P = 0.088). The number period covered (aural: F = 0.070; df = 1, 3; P = of calling birds detected in the dry zone was sig- 0.808; visual: F = 0.031; df = 1, 3; P = 0.868; nificantly different from the moist and wet zones total: F = 0.149; df = 1, 3; P = 0.725). (F = 55.875; df = 2, 24; P -c 0.0001). Similarly, the number of visual detections in the dry zone SCALY-NAPED PIGEONS was significantly different from the moist and wet A significant peak of Scaly-naped Pigeonscalling zones (F = 79.908; df = 2, 3; P = 0.002). Visual was detected during spring in the life zones for detections in the dry and moist zones appeared the 3 years (F= 23.862; df = 3, 108; P < 0.000 1). to peak between June and November (F = 9.178; In the moist and wet zones, calling activity ex- df = 3,9; P = 0.004). Aural and visual detections tended from January until at least July, whereas of Common Ground-Doves in the PSUs of the calling activity from September to November wet zone were negligible during the sampling pe- was negligible. Visual detectionsin the moist and riod covered. Five Common Ground-Doves were wet zones peaked between June and October (F detectedafter conducting 720 point counts in the = 3.988; df = 3, 9; P = 0.046). Aural and visual wet zone. The seasonalpatterns of aural and vi- detections in the PSUs of the dry zone were neg- sual detections in the dry and moist zones re- ligible during the surveys. Scaly-naped Pigeons mained unchanged after two years of counts (au- were detected only twice after conducting 1,080 ral:F=2.012;df= 1,3;P=0.251;visual:F= point counts in the dry zone. 0.016; df = 1, 3; P = 0.907; total: F = 0.824; df The number of calling birds detected in the = 1, 3; P = 0.561). wet zone was significantly different from the moist When the call counts of Common Ground- and dry zones (F = 20.117; df = 2, 36; P < Doves and Zenaida Doves in the dry zone were 0.0001). Even when the dry zone was excluded pooled, May stood as the month of peak activity. from the analyses, the significant difference pre- The relative abundance of Common Ground- dominated (F = 5.757; df = 1, 24; P = 0.024). Doves appeared to be higher than the relative The calling activity in the wet zone was signifi- abundance of Zenaida Doves, but significance cantly different from the moist zone during spring was reached only during fall (F = 4.248; df = 1, (F = 10.286; df = 1, 74; P = 0.002) and ap- 45; P = 0.045). proached significance during winter (F = 3.532; df = 1, 74; P = 0.064) and summer (F = 3.138; RUDDY AND KEY WEST QUAIL-DOVES df = 1, 74; P = 0.08 1). The seasonalpatterns of Two years of data collection indicated that total aural and visual detections in the moist and wet detections(aural + visual) of quail-doves (pooled) zones remained unchanged during the sampling peaked during the spring-summer period (F = period covered (aural: F = 2.410; df = 1, 3; P = 3.7 17; df = 3, 72; P = 0.015). The number of 0.218; visual: F = 0.03 1; df = 1, 3; P = 0.870; total detections in the moist and wet zones were total: F = 1.636; df = 1, 3; P = 0.29 1). significantly different from the dry zone (F = When the call counts of Scaly-naped Pigeons 4.860; df = 2, 64; P = 0.017). However, the and Zenaida Doves in the moist zone were pooled, counts of the moist zone were significantly dif- April stood as the month of peak activity. The ferent from the wet zone during spring(F= 3.353; relative abundance of Zenaida Doves appeared df = 2,96; P = 0.039) whereasthey approached to be significantly higher than the relative abun- significance during summer (F = 2.577; df = 2, dance of Scaly-naped Pigeons during winter (F 96; P = 0.08 1). Detections of quail-doves in the = 6.188; df = 1, 49; P = 0.016), spring (F = dry zone were mainly restricted to the xerophytic 230 FRANK F. RIVERA-MILAN

(n = 15 principalsampling units)

Months (Jul. 1986&n. 1988) FIGURE 2. Columbidsheard calling and seenin the 15 PSUs of the threemajor life zonesof PuertoRico.

forest of Guanica (PSU 14). Quail-doves were MOURNING DOVES never detectedin the dry coastalhabitats of PSUs Total detections of Mourning Doves peaked 15,28,29, and 36. The seasonalpatterns ofaural during summer (F = 3.095; df = 3, 72; P = and visual detections (pooled) remained un- 0.0322). Detections in the dry zone were signif- changed during the surveys (total: F = 2.659; df icantly different from the moist and wet zones = 1, 3; P = 0.361). (F = 6.311; df = 2, 24; P = 0.006). The counts WHITE-WINGED DOVES of the dry zone were significantly different from the moist zone during spring (F = 3.127; df = 2, Total detections of White-winged Doves peaked 77; P = 0.049), summer (F = 6.186; df = 2, 77; during summer (F = 3.095; df = 3, 72; P = P = 0.002) and fall (F = 5.4 15; df = 2, 77; P = 0.032). Relative abundance estimates in the dry 0.006). As in the case of White-winged Doves, zone were significantly different from the moist detections in the PSUs of the moist zone were and wet zones (F = 6.3 11; df = 2,24; P = 0.006). mainly restricted to agricultural areas (e.g., PSU The counts of the dry zone were significantly 32), whereas detections in the PSUs of the wet (F = different from the moist zone during spring zone were negligible. The seasonal patterns of 3.127;df= 2,77; P= 0.049) summer(F= 6.186; aural and visual detections (pooled) remained df = 2, 77; P = 0.002), and fall (F = 5.415; df = unchanged during the surveys (total: F = 3.176; 2, 77; P = 0.006). Detections in the PSUs of the df = 1, 3; P = 0.1728). moist zone were mainly restricted to agricultural areas (e.g., PSU 32) whereas detections in the COLUMBIDS PSUs of the wet zone were negligible. The seasonal patterns of aural and visual detections The relative abundance of all columbids in the (pooled) remained unchanged during the sam- three major life zones ofthe island peaked during pling period covered (total: F = 0.001; df = 1, the spring-summer period (March-August: F = 3; P = 0.976). 21.575; df = 3, 108; P < 0.0001; Fig. 2). Aural COLUMBIDS IN PUERTO RICO 231

2.5 (n = 15 principal sampling units)

+- Jul. 86-Jun. 87 * Jul. 87-Jun. 88

(paired t-test; P = 0.279; df = 1 1; overall) I I I

(n = 15 principal sampling units)

3.2 -

2.8 -

Q Jul. 86-Jun. 87 -+ Jul. 87-Jun. 88 2.6 -

(paired t-test; P = 0.297; df = 1 1 ; overall)

I I I I WINTER SPRING SUMMER FALL

Seas0 ns FIGURE 3. Seasonalpatterns of aural (A) and total (B) detectionsof all columbids in the wet, moist, and dry zones (log-transformed). detections peaked in April, before the peak of t-test; P < 0.05). This significant difference ap- nesting activity in the life zones (May-June), peared to be the result of an increasein the num- whereas visual detections remained high be- ber and size of flocks of White-winged Doves tween April and August. The second(July 1986- and Mourning Doves detectedin agricultural ar- June 1987) and third (July 1987-June 1988) year eas of the dry and moist zones (Rivera-Milan of study showed similar patterns of relative 1990). abundance(aura1:F= 1.660; df= 1,3; P= 0.288; The seasonalcalling patterns ofzenaida Doves total: F = 1.593; df = 1, 3; P = 0.296; Fig. 3). and Common Ground-Doves in the dry zone The third year, however, resulted in higher sum- were obscured when the counts of the life zones mer counts than the secondyear of study (paired were pooled (compare Figs. 2, 3 and 4). The 232 FRANK F. RIVERA-MILAN

0 Zenaida Doves (n = 5 principal sampling units) + Ground Doves

Months (Jun. 1986-J& 88) FIGURE 4. Calling patternsof ZenaidaDoves and CommonGround-Doves in the PSUsof the dry zone. calling patterns of Zenaida Doves and Common Common Ground-Doves is = 35 g, whereas the Ground-Doves in the PSUs of the dry zone were body mass of Zenaida Doves is x 150 g. Hence, correlated and appeared to be triggered by the according to the model devised by MacMillen rainfall bimodality (r = 0.7; P < 0.0001; rr = (1990), Common Ground-Doves should be more 0.6; P c 0.001; n = 24; Fig. 5). When the calling efficient than Zenaida Doves in deriving their activity of Zenaida Doves in PSUs 14 and 15 energy and water requirements from a single for- was pooled, a significant correlation with rainfall was detected (r, = 0.4; P < 0.025; and r. = 0.7; P < 0.001; n = 24, respectively). Similarly, when 25.4 the calling activity of Common Ground-Doves -D rain Gu6nica -+ rain CaboRojo in PSUs 14 and 15 was pooled, the correlation 20.3 = with rainfall was significant (r, = 0.5; P < 0.01 L n e and rs = 0.5; P < 0.01; n = 24, respectively). f‘ 15.2 Common Ground-Doves, however, are probably less constrained than Zenaida Doves by rainfall seasonality. Common Ground-Doves forage z mainly on annual and perennial grassseeds (e.g., E 2.2 Guinea grass [Panicum maximum]), whereas Zenaida Doves occupy both granivore and fru- givore guilds. Frugivores can be expected to be 0 I .,.,.,.,.1.,.,.,.,.,‘r highly sensitive to rainfall seasonality in xeric Jan Feb Mar Apr May Jun Jul Aug Sep Ott NW Dee environments, since fruits tend to be patchily- f-lonths distributed in spaceand time, and can be in short- FIGURE 5. Monthly rainfall in the xerophyticforest supply during prolonged periods of dryness of Guanica(PSU 14)and the dry coastalforest of Cabo (Faaborg 1982). Moreover, the body mass of Rojo (PSU 15).

234 FRANK F. RIVERA-MILAN

g e 0 0 2.5 5.1 7.6 10.2 12.7 15.2

DFN (cm of rainfall) FIGURE 6. Number of ZenaidaDoves calling vs. departuresfrom normal (DFN) rainfall duringthe first six monthsof the year (1985-1988) in PSUs 14 and 15 (n = 18). activity of Scaly-naped Pigeons (y = -0.502 + wet and moist zones peaked during spring and 0.027x, + 0.02~~ - 0.01~~ + 0.018x,; r* = 0.76; then declined after early-summer (June). How- P = 0.0249; df = 11) and Zenaida Doves (y = ever, the seasonal calling patterns of Zenaida -5.181 + 0.016x, + 0.07~~ - 0.002x, + 0.01 lx,; Doves and Common Ground-Doves in the dry r2 = 0.805; P = 0.0126; df = 1 l), respectively, zone were bimodal, with peaks occurring during in Rio Abajo forest. spring and fall. Fall aural detections of White- A significant relationship was found between winged Doves and Mourning Doves in the dry the calling activity of Scaly-naped Pigeons and zone were negligible during the sampling period Zenaida Doves (pooled) and the availability of covered. The second and third week of July fruits of key tree species in the riparian forest marked the beginning of the flocking period, and bordering Carite Lake and the secondarygrowth aural detections declined while visual detections forestsof Cidra and Comerio (pooled) (y = 0.600 became predominant. Zenaida Doves, Common + 0.649x; rz = 0.318; P = 0.054; df = 11). The Ground-Doves, White-winged Doves, and calling activity of Scaly-naped Pigeons in PSU Mourning Doves formed foraging aggregations 12 was associatedwith the availability of fruits in agricultural areas of the dry and moist zones, of S. jumbos in the strip-transectsbordering Car- whereas flocks of Scaly-naped Pigeons were de- ite Lake (y = 1.579 + 4.822x; r2 = 0.774; P < tectable in wooded habitats of the interior and 0.001; df = 11). A significant relationship was coast. A flocking period was not discernible for found between the calling activity of Scaly-naped the quail-doves. Quail-doves were usually de- Pigeons and Zenaida Doves (pooled) and the tected while calling (62%) or when foraging on availability of fruits of S. jumbos in the PSUs of the ground in small groups(5 11 birds) with vari- the wet and moist zones (wet: y = 0.93 + 0.03 lx; able degreesof cohesiveness(Rivera-Milan, un- rz = 0.803; P = 0.00 1; moist: y = 1.379 + 0.0 18x; publ. data). Overall, the flocking period of co- rz = 0.492; P = 0.011; df = 11, respectively). lumbids coincided with the termination of a significant portion of their reproductive pulse in DISCUSSION the island (Rivera-Milan 1990). DISTRIBUTION AND RELATIVE The seasonal patterns of relative abundance ABUNDANCE PATTERNS were detectable at local (PSUs) as well as at re- The counts of columbids resulted in repeatable gional (life zones) scalesof sampling, regardless patterns. The calling activity of columbids in the of the observer involved and the counting tech- COLUMBIDS IN PUERTO RICO 235

1 .8

2 3 4 5 6 7

Fruit availability FIGURE 7. Number of Scaly-napedPigeons calling vs. fruit availability of immatureand maturefruits of key treespecies (S. jambos, R. borinquena,F. citrifolia, C. schreberiana,S. torvum, Z. martinicense,E. poeppigiana, G. spicuta,and M. foetidissimum;k ~ 1 = 8) in the subtropicalwet forestof Rio Abajo from 1984 to 1985 (n = 12). Data sources:Hammerson et al., 1984;Cardona et al.. 1986. nique applied (see, for example, Sedgwick 1976, variable detection capacities are conducting the 1977; Hammerson et al. 1983, 1984; Moreno- counts (see Bart 1985). A plausible alternative Brillon et al. 1986; Rivera-Milan, in prep.). The seems to be the calibration of unfixed and fixed detected patterns are not considered to be either counts via double-sampling (Cochran 1977, the result of sampling error (e.g., due to insuffi- Eberhardt and Simmons 1987, Pollock and Ken- cient replication per life zone) or of statistical dall 1987, Rivera-Milan 1990). Counting stan- artifacts (e.g., created when pooling counts of dards should be as rigorous as possible to justify populations with independent dynamics). How- broad comparisons of speciesabundances. The ever, relative abundance estimates based on au- seasonalpatterns of relative abundance detected ral and visual detections are commonly assumed in the life zones probably representthe most gen- to be valid indicators of population density eral and repeatable aspect of the population dy- changesat different spatial and temporal scales namics of columbids in the island. Management of sampling (but see Rotella and Ratti 1986). and research decisions concerning columbid Seasonal estimates of relative abundance need populations should be judged initially at a life to be adjusted for effective area of detection, es- zone and seasonalscale of resolution (see Wiens pecially when comparisons of different speciesin et al. 1986). heterogeneoushabitats are of interest (seeVerner 1985). The PSUs under study, for example, are COMPONENTS OF THE ENVIRONMENT OF located in habitats with marked differences in COLUMBIDS vegetation structure and composition, weather, On average, rainfall in PSUs 14 and 15 peaked terrain, and urbanization levels among other in May and from October to November. The dry things. If detectiondistances remain unfixed, even seasonnormally extended from December until call counts of conspicuousspecies can be affected at least April. Therefore, the rains of May were because of problems of acousticsand unknown expected to provide the most important envi- rates of sound attenuation in different habitats ronmental stimuli for the calling activity and re- (Emlen and DeJong 198 1, Richards 198 1, DeJong production of columbids in the dry zone. The and Emlen 1985, Waide and Narins 1988). The first peak of rain (May) marks the end of the dry problem is complicated when observers with season,and normally is followed by a period of 236 FRANK F. RIVERA-MILAN increasing soil moisture and primary productiv- fall (Rivera-Milan, in prep.). The effectsof rain- ity (Lug0 et al. 1978). On the other hand, the fall fall on the reproduction of Zenaida Doves are rainy seasonis followed by a period of extended detectable via the call count survey from April drynesscharacterized by a deficit of soil moisture to June of the next two years, unless dramatic and a decline in vegetation cover and primary changes in the environment occur (e.g., hurri- productivity (Lug0 et al. 1978; Rivera-Milan, canes; Rivera-Milan, in prep.). pers. observ.). Lugo et al. (1978) reported a pos- Frugivores, and to a lesser extent granivores, itive correlation between rainfall and soil mois- should be sensitive to rainfall seasonalityin xeric ture availability, and a concomitant positive cor- environments (see, for example, Faaborg 1982). relation between soil moisture and grossprimary However, rainfall as a proximate factor probably productivity in the xerophytic forest of Guanica is less important than fruiting phenology of key (PSU 14). The increase in primary productivity tree species in mesic environments. The rela- should be linked with an increase in secondary tionship of rainfall and fruiting phenology in me- productivity (e.g., biomass of herbivores). The sic environments is complex and species-specif- combination of these forms of productivity ic; but, in general,the period of highestavailability should enhance the reproductive performance of of fruits appearsto occur from December to June, birds in xeric environments (but seeRivera-Mil- whereasthe period of lowest availability appears an 1990, for complexities arising from the con- to occur from July to November (see, for ex- comitant increasein the density of predatorssuch ample, Cardona et al. 1986). The calling activity as Pearly-eyed Thrashers [Murgurops fuscatus] of Scaly-naped Pigeons and Zenaida Doves in and Red-legged Thrushes [ Turdus plumbeus] the moist and wet zones appeared to be triggered among others). by the availability of fruits of key tree species Faaborg and Arendt (in press) hypothesized such as S. jambos. However, detailed studies that the reproductive successof birds in the xero- about the foraging ecologyof columbids in Puer- phytic forest of Guanica was negatively corre- to Rico are presently lacking (see, for example, lated either with positive or negative rainfall ex- Bancroft et al. 1990). Although food availability tremes (DFN values) occurring during the first is often cited as a factor of ultimate importance part of the year. However, they noted that after in the reproduction of birds (Andrewartha and 15 years of study, DIN values were dominated Birch 1954,1984; Immelmann 1971; Lack 1976), by periods of severe drought, and suggestedthat it seems reasonable to hypothesize that the rel- the negative correlations found represented the ative abundance patterns of columbids have not impact of negative extremeson reproductive suc- been molded solely by the availability of primary cess.Above normal rainfall values (i.e., positive productivity in any of the life zones. In the moist extremes) were positively correlated with the and wet zones, columbids find fruits available calling activity of Zenaida Doves during the first on a year-round basis, especially in natural and part of the year in PSUs 14 and 15. Time lags human-induced forest gaps (e.g., R. borinquena, of one and two years were detectable in the re- C. schreberiana, and Zanthoxylum martinicense, sponsivenessof calling Zenaida Doves to normal among others). Moreover, during prolonged pe- and above normal rainfall values. Hence, my riods of dryness and food scarcity columbids data support the hypothesis of Faaborg and might easily disperse to neighboring forests in Arendt (in press)that the rainfall of the first six the ecotone of the dry and moist zones (Rivera- months ofthe year representsthe most important Milan, ongoing research). environmental component influencing the repro- The distribution and relative abundance pat- duction of birds in the Guanica xerophytic forest. terns of columbids probably are under the influ- The calling and nesting activity of Zenaida Doves ence of several components of the environment should be negatively correlated with below nor- acting in a density-dependent or -independent mal rainfall values occurring during the first part mode. The relationship between biotic and cli- of the year in the xerophytic and dry coastal matological variables might be complex, and the forestsof the dry zone. Prolonged periods of dry- observed patterns can be erroneously attributed ness during the first part of the year might affect to the effectsof the most evident or measurable. the abundance of columbids in the dry zone. The Population fluctuations at local scalesmight be reproductive performance of Zcnaida Doves is influenced by predators, parasites, diseases,and enhanced during years with above normal rain- marginal deficiencies and imbalances of nutri- COLUMBIDS IN PUERTO RICO 237 ents among other things (Rivera-Milti 1990). BIRDSEY,R. A., AND P. L. WEAVER. 1982. The forest Nevertheless, it seems reasonable to conclude resourcesof Puerto Rico. U.S. Dep. Agric. For. that rainfall in the dry zone and fruiting phe- Serv. Resour. Bull., SO-85, South For. Exp. Stn., New Orleans, LA. nology of key tree speciesin the moist and wet CARDONA, J., M. RMXA, M. VAzqu~z, AND C. R. zones represent the most important environ- LABOY. 1986. Availability of food resourcesfor mental components influencing the seasonalpat- thePuerto Rican Parrot and the Puerto Rican Plain terns of relative abundance of columbids in the Pigeonin Rio Abajo Forest.Project W- 10 (ES- 1). ScientificResearch Area. Puerto Rico Dept. Nat. island. Res. COCHRAN,W. G. 1977. Sampling techniques.3rd ed. ACKNOWLEDGMENTS John Wiley and Sons, New York. COLLAZO-ALGAR~N,J., G. BONILLA, D. RAMOS, E. This investigation was performed in partial fulfillment RAMOS,AND A. ORTIZ. 1985. Project W-l l-4. of the requirements for the degree of Ph.D. from the Annual Performance Report. Scientific Research University of Maryland at CollegePark (UMCP). I am Area. Puerto Rico Dept. Nat. Res. grateful to the following people for their suggestions DAY, R. W., AND G. P. QUINN. 1989. Comparisons and recommendations concerning the material pre- of treatments after an analysisof variance in ecol- sentedhere: Dale Bonar, Julio Cardona, Jaime Collazo, ogy. Ecol. Monogr. 59:433-463. David D. Dolton. John Faaborp.Paul H. Geissler. Ste- DEToNG,M. J., AND J. T. EMLEN. 1985. The shape phen Leatherman, Robert E. Menzer, Donald H. &les- of auditory detection function and its implications sersmith, Kenneth H. Pollock, Fred Schaffner, and for songbirdcensusing. J. Field Omithol. 56:2 13- James W. Wiley. I extend my gratitude to the crew of 320. - Project W- 11-PRDNR for their interest and assistance DESANTE,D. F., AND G. R. GEUPEL. 1987. Landbird at the field, including Gilbert Bonilla, Anastacio Ortiz, productivity in central coastal California: the re- Jest PagBn, David Ramos, and Edgardo Ramos. J. lationship of annual rainfall, and a reproductive Cardona provided constant accessto the data about failure in 1986. Condor 89:636-653. fruiting phenologyof trees in the subtropicalwet forest DOLTON.D. D. 1985. Mourning Dove breedingDOD- of Rio Abajo. Miguel Canalsprovided the rainfall data ulation status.Adm. Rep. Dept. Int. U.S. Fishand of the xerophytic forest of GuBnica,and J. Collazo, Sue Wildl. Serv. Rice, and F. Schafier provided the rainfall data of the EBERHARD-~,L. L., AND M. A. SIMMONS. 1987. Cal- dry coastal forest of Cabo Rojo. Finally, I would like ibration ofpopulation indices by double sampling. to thank Jest Vivaldi, who provided constant support J. Wildl. Manage. 51:665-675.. as coordinator of the Pittman-Robertson Funds grant- EMLEN.J. T.. ANDJ. DEToNG. 1981. The annlication__ in-aid to the Section of Terrestrial Ecology, Scientific of songdetection threshold distance to censusop- Research Area, Puerto Rico Department of Natural erations, p. 346-352. In C. J. Ralph and J. M. Resources;and George Proctor for helping in the iden- Scott[eds.], Estimating numbers ofterrestrial birds. tification of plants, seeds, and fruits whenever it was Stud. Avian Biol. No. 6. Cooper Omithol. Sot. requested. From May 1984 to February 1987, I also Allen Press, Inc., Lawrence, KS. received supportas a researchassistant from the Grad- EWEL.J. J.. ANDJ. L. WHITMORE. 1973. The ecoloa- uate Program in Marine and Estuarine Environmental ical life zones of Puerto Rico and U.S. Virgin Ii- Sciences(MEES) and the Agricultural and Life Sciences lands. For. Serv. Res. Paper ITF-18. Division at UMCP. FAAEXORG,J. 1982. Avian population fluctuations during drought conditions in Puerto Rico. Wilson Bull. 94:20-30. LITERATURE CITED FAAEXORG,J., AND W. J. ARENDT. In press. Rainfall correlates and bird population fluctuations in a ANDREWAR-~HA,H. G., AND L. C. BIRCH. 1954. The Puerto Rican dry forest: a 15 year study. Omitol. distribution and abundance of animals. Univ. of Carib. Chicago Press, Chicago. FAABORG,J., W. J. ARENDT,AND M. S. KAISER. 1984. ANDREWARTHA,H. G., AND L. C. BIRCH. 1984. The Rainfall correlatesof bird population fluctuations ecologicalweb. Univ. of Chicago Press, Chicago. in a Puerto Rican dry forest: a nine year study. BANCROF~,G. T., R. BOWMAN.R. J. SAWICKI.AND A. Wilson Bull. 96~575-593. M. STRONG. 1990. Relationship between the reproductive ecology of the White-crowned Pigeon GOODWIN,D. 1983. Pigeonsand doves of the world. and the fruiting phenology of tropical hardwood 3rd ed. Cornell Univ. Press, Ithaca, NY. hammock trees. Florida Game Fresh Water Fish GUREVITCH,J., ANDS. T. CHESTER,JR. 1986. Analysis Comm., Nongame Wild]. Program Tech. Rep. ofrepeated measuresexperiments. Ecology 67:25 l- BART,J. 1985. Causesof recording errors in singing 255. bird surveys. Wilson Bull. 97: 16 l-172. HAMMERSON,J., G. BONILLA,D. RAMOS,E. RAMOS, BEAL, K. G., AND H. J. KHAMIS. 1990. Statistical AND F. RIVERA. 1983. Project W-l l-2. Annual analysis of a problem data set: correlated obser- Performance Report. Scientific Research Area. vations. Condor 92:248-25 1. Puerto Rico Dept. Nat. Res. BEAL,K. G., ANDH. J. KHAMIS. 1991. Reply to Quinn HAMMERSON,J., G. BONILLA,D. RAMOS,E. RAMOS, and Keough. Condor 93:202-203. AND F. RIVERA. 1984. Project W-l l-3. Annual 238 FRANK F. RIVERA-MILAN

Performance Report. Scientific Research Area. censusesof singing birds, p. 297-300. In C. J. Puerto Rico Dept. Nat. Res. Ralph and J. M. Scott teds.], Estimating the num- IMMELMANN, K. 1971. Ecologicalaspects of periodic bers of terrestrial birds. Stud. Avian Biol. No. 6. reproduction, p. 341-389. In D. S. Famer and J. Cooper Omithol. Sot. Allen Press,Inc., Lawrence, R. King [eds.],Avian Biol. Vol. 1. Academic Press, KS. New York. RIVERA-MIL.&N,F. F. 1990. Distribution and abun- KLEINBAUM,D. G., ANDL. L. KUPPER. 1978. Applied dance of columbids in Puerto Rico. Ph.D.diss. regressionanalysis and other multivariate meth- Univ. of Maryland, College Park, MD. ods. Duxbury Press, Boston, MA. RIVERA-MIL& F. F., G. BONILLA,D. RAMPS,E. RAMOS, LACK, D. 1976. Island biology. Univ. of California AND A. ORTIZ. 1990. Project W-l 1. Final Per- Press, Berkeley and Los Angeles. formance Report. Puerto Rico Dept. Nat. Res. LUGO,A. E., J. A. GONZALEZ-LIBOY,B. CINTR~N,AND ROTELLA,J. J., AND J. T. RA-~TI. 1986. Test of a K. DUGGER. 1978. Structure, productivity, and critical density index assumption:a casestudy with transpiration of a subtropical dry forest. Biotro- gray partridge. J. Wild]. Manage. 50:532-539. pica 10:278-291. SEDGWICK,J. A. 1976. Project W-8-17. Annual Per- MACMILLEN,R. E. 1990. Water economy of graniv- formance Report. Scientific ResearchArea. Puerto orous birds: a predictive model. Condor 92:379- Rico Dept. Nat. Res. 392. SEDGWICK, J. A. 1977. Project W-8-18. Annual Per- MORENO-BRIL~~N,J., G. BONILLA, D. RAMOS, E. formance Report. ScientificResearch Area. Puerto RAMOS,AND A. ORTU. 1986. Apuntes sobre la Rico Dept. Nat. Res. distribucibn, abundancia relativa, y fenologia re- SOKAL,R. R., ANDF. J. ROHLF. 1981. Biometry. 2nd productiva de la T6rtola Cardosantera (Zen&u ed. W. H. Freeman and Co., New York. aurita) y la Paloma Turca (Columba squamosa) VERNER,J. 1985. Assessmentofcounting techniques, en Puerto Rico. Omitol. Carib. 2:35-42. D. 247-302. In R. F. Johnston led.1. Current or- P&~-RIvERA, R. A., ANDJ. COLLAZQ-ALG&. 1976. nithology. Vol. 2. Plenum Press:N&v York. Distribucibn geogrifica, habitos alimentarios, y WAIDE, R. B., AND P. M. NARINS. 1988. Tropical competenciapor alimentos de la Paloma Sabanera forest bird counts and the effect of sound attenu- (Columba inornutuwetmorei) de Puerto Rico. Sci- ation. Auk 105:296-302. ence-Ciencia 3:52-55. WIENS,J. A., J. F. ADICO~, T. J. CASE,AND J. DIAMOND. POLLOCK,K. H., AND W. L. RANDALL. 1987. Visi- 1986. Overview: the importance of spatial and bility bias in aerial surveys:a review of estimation temporal scalein ecologicalinvestigations, p. 145- procedures.J. Wildl. Manage. 5 1:502-509. 153. In J. Diamond and T. J. Case [eds.], Com- QUINN, G. P., AND M. J. KEOUGH. 1991. Repeated munity ecology. Harper and Row, New York. measuresanalysis of variance: a comment on Beal WILEY, J. W., AND B. N. WILEY. 1979. The biology and Khamis (1990). Condor 93:200-20 1. of the White-crowned Pigeon. Wildl. Monogr. No. RAFFAELE,H. 1989. A guide to the birds of Puerto 64. Rico and the Virgin Islands. 2nd ed. Princeton WINER, B. J. 1971. Statistical principles in experi- Univ. Press, Princeton, NJ. mental design. 2nd ed. McGraw-Hill, New York. RICHARDS,D. G. 1981. Environmental acousticsand