Ecology and Conservation of Grassland Birds of the Western

Studies in Avian Biology No. 19:289-299, 1999.

WINTER ECOLOGY, BEHAVIOR, AND CONSERVATION NEEDS OF DICKCISSELS IN VENEZUELA

GIANFRANCO D. BASILI AND STANLEY A. TEMPLE

Abstract. Dickcissels (S&a americana) were studied in 1990-1993, and briefly in 1995, during the nonbreeding season in agricultural regions of the Venezuelan llanos. These llanos comprise the Dick- cissels’ core wintering area, and birds occurred there for more than seven months, September through April. Dickcissels were gregarious, feeding and roosting together in large flocks. We found that sugarcane (.Saccharum sp.) fields were the preferred roosting habitat, and we estimated that individual roost sizes could reach three million birds. Dickcissels were mainly granivores in winter, and while in the llanos they fed on wild grass seeds and agricultural crops, especially sorghum (Sorghum vulgare) and rice (Otyza sativa). Since the 1950s increases in cereal agriculture in Venezuela have provided Dickcissels with a superabundant food supply, yet the Breeding Bird Survey has indicated a 40 percent population re- duction in North America since monitoring efforts were initiated in 1966. We think this decline has likely resulted from high overwinter mortality induced by lethal control operations. In many areas of Venezuela, farmers consider Dickcissels to be agricultural pests. Although most farmers use nonlethal controls to mitigate damage, some intentionally kill Dickcissels with agricultural chemicals. The Dick- cissels’ highly gregarious nature and small geographic range make the entire population vulnerable to catastrophic mortality. We propose that further efforts to prevent declines, and possibly to restore Dickcissel populations to historical levels, be aimed at reducing lethal control on the species ’ central wintering grounds in Venezuela.

ECOLOGIA, COMPORTAMIENTO Y NECESIDADES DE CONSERVACIGN DEL ARROCERO AMERICAN0 DURANTE EL INVIERNO EN VENEZUELA Sinopsis. Estudiamos 10s Arroceros Americanos (Spiza americana) entre 1990-1993, y brevemente en 1995, durante la estacion no-reproductiva en regiones agrfcolas de 10s llanos de Venezuela. Estos llanos contienen el corazon de1 area invemal de1 Arrocero Americano, y las aves estuvieron alli por mas de siete meses, de septiembre hasta abril. Los Arroceros Americanos eran gregarios, alimentandose y descansando juntos en grandes bandadas. Encontramos que 10s posaderos preferidos eran 10s campos de cafia de azucar (Saccharum sp.). y estimamos en tres millones de aves las agrupaciones de aves. Los Arroceros Americanos eran principalmente granivoros durante el invierno, y mientras estaban en 10s llanos se alimentaban con semillas de hierbas silvestres y cosechas agrfcolas, especialmente sorgo (Sorghum vulgare) y arroz (Olyza sativa). Desde 10s aAos 50, el aumento de la agricultura de cereales en Venezuela ha suministrado provi- siones alimenticias excesivamente abundantes, pero el Breeding Bird Survey ha indicado una dismi- nucidn de poblacidn de un 40 por ciento en America de1 Norte desde que se iniciaron 10s esfuerzos de medicion en 1966. Creemos que es probable que esta disminucion sea el resultado de la alta mortalidad durante el invierno provocada por operaciones letales de control. En muchas areas de Venezuela, 10s agricultores consideran a 10s Arroceros Americanos coma pestes agricolas. Aunque la mayoria de 10s agricultores usa controles no letales para mitigar 10s dafios, algunos matan a 10s Arroceros Americanos intencionalmente con sustancias quimicas agrfcolas. El caracter gregario de1 Arrocero Americano y su pequefia extension geografica hacen que la poblacion entera sea vulnerable a la disminuci6n de1 Arrocero Americano y a una mortalidad catastr6fica. Proponemos que 10s futures esfuerzos para impedir declinaciones y posiblemente para restaurar sus poblaciones a niveles histbricos, Sean dirigidos a la reduccidn de1 control letal en 10s principales campos invernales de la especie en Venezuela.

Key Words: Dickcissel; feeding behavior; lethal poisoning; mortality; population decline; roosting behavior; Spiza americana; winter distribution.

Studying the population dynamics of nearctic- been well studied, relatively little is known neotropical migrants is complicated because im- about its ecology during the nonbreeding season. portant demographic events take place over huge This dearth of information confounds conser- geographic areas (Maurer and Villard 1996). vation efforts because it is impossible to imple- Most research on migrants, especially grassland ment successful management plans when uncer- species such as the Dickcissel (Spiza america- tainties remain concerning the species ’ geo- na), has taken place during the breeding season graphic distribution, behavioral ecology, and (Petit et al. 1995). Although the Dickcissel has natural history during more than half of its an-

289 290 STUDIES IN AVIAN BIOLOGY NO. 19

Winter rangeof the Dick&se1 Area in which mxt Dickcisselswinter

FIGURE 1. Primary winter range of the Dickcissel in South America.

nual cycle (Petit et al. 1995). Although we agree there include rice (Oryza sativa), sorghum (Sorghum that existing knowledge of neotropical migrants vulgare), and sugarcane (Saccharum sp.), but tobacco should be used for immediate conservation ac- (Nicotiana sp.), cotton (Gossypium sp.), tomatoes tion (Martin and Finch 1995), we remain cau- (Lycopersicon esculentum), sesame (Sesamum indi- cum), and sunflowers (Helicanthus sp.) are also culti- tious because our experience with Dickcissels, vated. Study sites in Guarico were in rice-growing ar- like those of Sherry and Holmes (1995) and eas south of the town of Calabozo. Rice is currently Temple (1995), suggests that taking action with- the only crop cultivated on a large scale in this area. out a complete understanding of the species ’ an- We located Dickcissels using road surveys. This nual cycle may lead to inappropriate and/or in- species ’ gregarious nature makes it conspicuous, es- efficient conservation strategies. pecially in the early morning and late afternoon when We reexamined the Dickcissels’ winter range flocks move between roosting and feeding areas. Noc- and identified patterns of abundance in this turnal roosts were located by following afternoon range. In this paper we provide information on flights. We plotted locations on topographic maps available from the Cartography Department of the Ve- natural history, diet and foraging, and habitat nezuelan Environmental Ministry (Ministerio de Am- use. We also describe conflicts between Dick- biente). Additional information on Dickcissel distri- cissels and Venezuelan farmers that may have bution in Venezuela and other parts of the winter range catastrophic consequences for the Dickcissel was obtained from the literature, personal communi- population. Finally, we suggest that conserva- cation with ornithologists and birdwatchers, and mu- tion efforts focus primarily on the Dickcissels’ seum skins from the Phelps Collection in Caracas winter range in Venezuela. (Venezuela), Field Museum of Chicago (Illinois, USA), American Museum of Natural History (New STUDY AREA AND METHODS York, USA), and Smithsonian Institution (Washington, Field research was conducted during the nonbreed- DC., USA). ing season in 1990-1993, and briefly in 1995, in the We estimated roost size by classifying incoming Venezuelan llanos (Fig. 1). We also reviewed museum flocks into size categories based on length, width, and specimens and literature and communicated with or- height (e.g., Newton 1972, Lindstriim 1989). We ob- nithologists, birdwatchers, and farmers in Trinidad, served foraging activity of males and females with a Venezuela, Colombia, Panama, Costa Rica, Nicaragua, window-mounted spotting scope (Litvaitis et al. 1994). Guatemala, and Mexico. The llanos that surround the Dickcissels are sexually dimorphic, and sex was easily Orinoco River in Venezuela and Colombia constitute determined by plumage (Pyle et al. 1987). We exam- the largest neotropical savanna north of the equator ined crop and gizzard contents of birds killed in con- (Sarmiento 1984). This region is characterized by a trol operations and mist-net sampling. The contents of rainy season from May through October and a dry sea- each sample were sorted by food item, oven-dried at son from November through April (Monasterio 1970). 60 C, and weighed. The proportion of the dry mass Our primary study sites were located in the grain-pro- represented by each food item was calculated (Litvaitis ducing regions of the states of Portuguesa, Cojedes, et al. 1994). and Gudrico. In Portuguesa and Cojedes, study sites We used focal sampling to document individual were located in the vicinity of the agriculturally pros- feeding behavior. The time between consecutive bites perous town of Acarigua. Dominant dry-season crops (handling time) for males and females was measured DICKCISSELS IN WINTER--Basili and Temple 291 for each food item. An individual’s handling time was Ranch0 Grande on 28 September 1996 (M. Len- calculated by measuring the time interval between the tino, pers. comm.). Thus, the most heavily used first and fourth bites, then dividing by three (the num- migration routes remain unknown. ber of seeds handled). These data were then aggregated Relatively few Dickcissels occur outside the and averaged by sex and food item. Fourteen individuals (12 males and 2 females) were core wintering area in Venezuela. A small flock fitted with 1.0-g radio transmitters attached with a har- of about 130 birds was reported near the mouth ness (Rappole and Tipton 1991). Because the range of of Rio Tocuyo, Falcbn, on 26 January 1995 (S. transmitters was approximately 1 km, these 14 birds Hilty, pers. comm.), and a group of 12 birds was were most easily detected at nocturnal roosts. Crop- observed in secondary old-field habitat during production data were obtained from the annual agri- the third week of January 1995 in the Cafio Col- cultural statistics (Anuario Estadistico Agropecuario) orado area east of Maturin, Monagas (C. Rodner, published by Venezuela’s Agriculture Ministry (Minis- pers. comm.). terio de Agricultura y Cria). We made a concerted effort to communicate with The llanos extend into Colombia, and some farmers throughout the project. Farmers not only pro- Dickcissels also winter there. For example, spec- vided accurate information on Dickcissel movements imens were collected in the department of Meta but were also candid about their ongoing struggle to (Los Micas, 28 February 1957), and Hilty and reduce damage caused by the birds. This communi- Brown (1986:642) refer to Dickcissels as a “location was essential to our work, and we recommend cally common winter resident but erratic in that it be an integral component of future Dickcissel many places.” Fretwell and Shane (1975) sur- research and conservation efforts in the region. veyed parts of the Colombian llanos near Villa- RESULTS vicencio but found few individuals. When information about Dickcissel damage to rice and sor- DICKCISSELMIGRATION AND DISTRIBUTIONIN ghum in Venezuela was presented to a rice- CENTRAL AND SOUTH AMERICA: A REVIEW growing association and rice experimental Dickcissels begin arriving in Venezuela in station in the Colombian llanos in 1995, audi- September (Femandez-Yepez 1945, Phelps and ences were surprised to learn that their Vene- Phelps 1963), with the largest numbers occur- zuelan counterparts were confronted with such ring from November through April. This 2-mo huge flocks (J. Botero, pers. comm.). These lag between first arrivals and peak numbers ap- farmers assured us that Dickcissels are currently pears to be the result of some birds spending not a problem for agriculture in Colombia, con- September and October in agricultural regions of firming our belief that central Venezuela is the Central America. Slud (1964) reported that be- main wintering area for the species. Museum ginning in September “clouds” of Dickcissels specimens collected between 1870 and 1950, depredated rice fields in the lower Pacific north- however, reveal Dickcissels wintering through- west of Costa Rica. Depredations continued out Venezuelan grasslands, including the states through November, with the species diminishing of Zulia (Guayabo, 27 February 1908), Barinas “enormously” from December through April. (Santa Barbara, 26 January 1947), Apure (San There are a few instances of Dickcissels staying longer in Central America. One example is a Fernando, 13 November 1948), Anzoategui large flock of 250,000 birds observed between (Santa Rosa, 25 January 1939), and the federal 19 and 21 February 1992 on the Pacific Coast territory of Amazonas (Ctio Catamiapo, 8 Feb- of Guatemala, near the village of Montericco (C. ruary 1943). Robbins, pers. comm.). Orians and Paulson In Trinidad, ffrench (1967) studied three (1969) reported a flock of 700-800 Dickcissels roosts from 1959 to 1966. Birds generally ar- on 10 December 1966 near Ctias in the Gua- rived in Trinidad by December and departed by nacaste province of Costa Rica, and Zimmerman mid-April. Fretwell (1986:213) stated that Dick- (1965a, b) observed wintering flocks in Panama cissels begin to leave central Venezuela in De- in January and February 1961. cember and January and arrive in “Trinidad to Museum specimens collected during migra- the east and Panama to the west by mid-Janu- tion indicate that Dickcissels enter and leave the ary.” Birds arriving in Trinidad undoubtedly do Venezuelan llanos via valleys in the Andes of so via Venezuela (ffrench 1967). The origins of western Venezuela (state of TBchira) and the January birds in Panama are less certain and do Coastal Corridor range of northern Venezuela not necessarily indicate a December movement (Henri Pittier National Park). Meyer de west from Venezuela. Schauensee and Phelps (1978) observed 1,000 In spring, Orians and Paulson (1967) reported Dickcissels per hour crossing at Ranch0 Grande thousands of Dickcissels migrating north in in Henri Pittier National Park in October. More Guanacaste, Costa Rica, between 27 and 30 recent studies (1990-1996) of fall migration March 1967, and thousands of individuals were yielded one Dickcissel captured in a mist net at observed flying north past the airport in Mana- 292 STUDIES IN AVIAN BIOLOGY NO. 19 gua, Nicaragua, between 21 and 27 March 1996 TABLE 1. AVERAGE PERCENT (BY DRY MASS) AND (R. Pumell, pers. comm.). PERCENT OCCURRENCEOFFOOD ITEMS INCROPS AND GIZ- ZARDSOF MALE (N = 35) AND FEMALE(N = 26) DICK- DICKCISSEL DISTRIBUTION IN VENEZUELA CISSELSIN MARCH 1992 IN THE VENEZUELAN LLANOS The largest Dickcissel concentrations we Average% % OCC"rrenCe found in Venezuela (l-6 million) were near the per cropa in cropsb agricultural cities of Acarigua and Calabozo. Food type Males F.XNkS Males Females This confirms that the Venezuelan llanos constitute the species ’ main wintering area. Smaller Plant matter populations are known to occur sporadically Rice 50 46 54 58 throughout Venezuela and in agricultural regions Sorghum 28 26 46 50 on Trinidad (< 50 km north of the Venezuelan Wild grass seedsC 16 16 49 58 Unidentified plant coast). matter 5 11 89 100 We found few Dickcissels outside the core wintering area in Venezuela. Our surveys of Animal matter southcentral Gutico, Apure, and Barinas in Insect parts 0.04 0.04 14 27 March 1991, March 1993, and April 1995 re- Spiders 0.01 0 6 0 vealed only small flocks of Dickcissels (loo- Grit 0.28 0.59 71 81

700) in rice fields near Sabaneta, Barinas. We aProportion of food items in males and females wac not significantly did not observe Dickcissels in regions dominat- different: f-tee usmg Bonferonni adjustments. ed by cattle ranching. b Frequency of occurrence of food items was not significantly different between males and females; C* = 2.516, df = 6, P > 0.8. Dickcissels remained in Venezuela after De- c Category comprised common grasses Orizya larifolia andRonoboellia cember. A survey between 24 January and 20 cochinchmensis. Three less common seed types were found in the crop and gizzard contents and remain unxdentified; they are accounted for m February 1993 revealed a population of about 6 the category “Umdentified plant matter.” million birds distributed among five roosts. Sim- ilarly large numbers were present during these months in 1991 and 1993. Dickcissels occurred in Venezuela for more than 7 mo, and their stay coincided with Vene- mum rugosum. Crop and gizzard contents of 61 zuelas’ dry season. In late March and early Dickcissels collected in March 1992 confirmed April, birds began to depart for temperate breed- the species ’ granivorous diet; 99% of the dry ing grounds. Large numbers remained in Vene- mass was seeds (Table 1). Rice was the most zuela until the third week in April, however, as abundant food item, followed by sorghum and evidenced by a roost near Caiio Seco, Portugue- wild grasses (Rottoboellia cochinchinensis and sa, that had an estimated 2,950,OOO birds on 15 Orizya latifolia). The remains of insects and spi- April 199.5. This was the largest roost we en- ders were found in 23% of all samples; however, countered during our study. By the end of April, their combined dry mass comprised less than when the rains usually begin (Monasterio 1970), 0.05% of the total sample (Table 1). There were most Dickcissels had migrated north. For ex- no statistically significant differences between ample, the Ctio Seco roost had nearly 3 million crop and gizzard contents (percent dry mass of birds on 15 April 1995 but only 20,000 on 24 food items and percent food item occurrence) of April 1995. A sample of 40 birds caught there males and females. on 25 April 1995 was heavily biased toward fe- FORAGING males (79%), indicating that males leave Vene- zuela before females. Our latest records included Dickcissels foraged in flocks that ranged in two males, one singing, at Pozo Blanco, Acari- size from a dozen individuals to enormous ag- gua, on 15 May 1995. gregations of several hundred thousand. The birds took seeds directly from ripening stalks or DIET from waste grain remaining after harvest (Fret- Although insects are an important component well 1986; G. Basili, pers. ohs.). It was unclear of the Dickcissels’ diet during the breeding sea- what percentage of the rice and sorghum con- son (Gross 1921), the species becomes graniv- sisted of waste grain, but the largest feeding ag- orous on its wintering grounds (Slud 1964, gregations (> 250,000 birds) were observed in ffrench 1967, Meyer de Schauensee and Phelps recently harvested rice fields. The largest aggre- 1978). We observed Dickcissels feeding mainly gations in sorghum fields were of similar size, on seeds. Direct observations of foraging indi- but they occurred in fields prior to harvest. Dick- viduals revealed five food items: rice, sorghum, cissels also fed in fallow agricultural fields and seeds from three wild grasses-Rottoboellia where wild grasses abounded (G. Basili, pers. cochinchinensis, Orizya latlfolia, and Ischae- obs.); aggregations in this habitat, however, were DICKCISSELS IN WINTER-Basili arzd Temple 293

TABLE 2. DICKCISSEL CROP AND GIZZARD CONTENTS TAKEN AT NOCTURNAL ROOSTS IN THE VENEZUELAN LLANOS, MARCH 1992

Number of individuals

Sorghum, RlCe Sorghum Grass Rice & R,ce & Sorghum rice, & R00st N OlllY Only Only sorghum grass & grass grass

Weiss 11 7 1 1 _ 1 1 Retejado 10 5 _ 1 _ 1 3 _ Pepe 15 _ 6 1 _ 1 5 2 Perez 2 10 9 _ _ _ 1 _ _

never as large as those we observed in rice and and dispersing to feed in different habitats, in- sorghum. cluding rice, sorghum, and fallow fields. Varia- A typical day of foraging is represented by tions in crop and gizzard contents in nocturnal the following account, recorded 26 January roosts confirmed this observation. A single food 1993. The site was a 20-ha recently harvested type predominated at only one of four roosts rice field 5 km from a nocturnal roost of 420,000 where we had 10 or more samples on a single birds. It was bordered on one side by a strip of evening (Table 2). Crop samples were not ob- trees along a dry stream bed. Old fields with tained in Gutico, but because sorghum was not shrubs, grasses, and forbs comprised the remain- grown there on a large scale, rice and seeds of ing habitat mosaic. The first group of birds wild grasses were the dominant food types. (1,000) arrived at 0717 and immediately began Foraging Dickcissels exploited certain areas feeding. Small groups continued to arrive until and then shifted to new feeding sites without 0755, bringing the total number of birds at the changing nocturnal roosts. Flight patterns of site to 5,000. During this time, many larger birds returning to roost often changed over time. groups passed high overhead and continued to For example, in one week most birds entered the more distant feeding areas. The greatest distance roost from the south, but the next week they between a roosting and feeding area was 20 km, arrived from the west. In December 199 1, morn- although Dickcissels have been reported to feed ing flights of Dickcissels passed directly over a up to 24 km from a roost (ffrench 1967). Birds ranch where we were stationed. For the first 3 fed until 1115, with groups (< 1,000 birds) pe- d, groups of less than 1,000 birds were present. riodically feeding in rice and then moving to rest On day four, numbers increased dramatically as in cover along the edges of the field. We ob- enormous columns of hundreds of thousands of served no feeding from 1115 to 1415, but small birds appeared, most flying past but thousands groups (< 30 birds) moved to different roost landing on the ranch to feed in old fields. Large sites bordering the rice. Feeding began again at numbers of Dickcissels persisted in the same 1415 as several thousand birds emerged from flight pattern for 4 d. On day 9, numbers con- the woods and settled into the rice. In the after- spicuously diminished, and by day 10 Dickcis- noon, groups (< 1,000 birds) moved periodical- sels were no longer present. We found that the ly between cover and rice, with the longest feed- same nocturnal roost was still active, but for- ing bouts lasting about 30 min. Feeding contin- aging sites had shifted just south of the ranch. ued intermittently until 1735, when birds from We observed feeding activity in a ripening more distant areas began flying overhead toward 120-ha sorghum field in March and April 1992. the nocturnal roost. On several occasions, strag- Birds were present for nearly 4 wk, with peak glers in high-flying groups spiraled down to the numbers exceeding 100,000 in the final week. rice field to feed. Finally, small groups (< 500 Even while the field was being harvested, Dick- birds) began to depart, and by 18 10 the feeding cissels dodged combines and fed on the remain- area was deserted. This schedule was typical in ing seeds. The day before harvesting, there were all habitats where Dickcissels fed. In the month 100,000 birds feeding in the field; the day after before their northward migration, however, harvesting was completed, the field was desert- Dickcissels became hyperphagic and spent more ed, even though waste grain was still abundant. time foraging as they attempted to build fat re- The nocturnal roost remained intact during this serves. period, but foraging flights were directed toward Birds that shared the same nocturnal roosts different areas. did not necessarily feed on the same food items We used focal sampling to document individ- during the day. In Portuguesa and Cojedes, we ual feeding behavior on four food types (rice, often observed flocks leaving nocturnal roosts sorghum, Rottoboellia cochinchinensis, and 294 STUDIES IN AVIAN BIOLOGY NO. 19

TABLE 3. MEAN TIME REQUIREDFOR FREE-RANGINGDICKCISSELS TO HANDLE ONE SEED OF FOUR GRASSSPECIES IN THE VENEZUELAN LLANOS,MARCH 1992

Meantime (xc) t SE(N) Females% slower Fooditem Males Females thanmales Rice 5.1 2 0.4(67) 6.8 C 0.6 (38) 33.3 Sorghum 6.2 -c 0.4 (57) 8.1 2 0.7 (25) 30.1 Orizya latifolia 5.9 k 0.7 (25) 6.7 2 0.5 (24) 13.6 Rottohoelliczcochinchinensis 15.6 2 0.8 (19) 16.2 k 1.4 (6) 3.9

Orizya latifolia). Birds consistently fed while area, at least 7 were used in multiple years. One perched on stalks, except when eating Rotto- was used for 4 yr (farm of R. Perez, near Cane boellia cochinchinensis, seeds of which were lo- Seco, Portuguesa) and another for 3 yr (farm of cated on the ground. A two-way analysis of var- J. Pinero, near Retejado, Cojedes). Three sug- iance (ANOVA) to evaluate the effects of sex arcane fields that were used by Dickcissels for and food item on bite times revealed that fe- only 1 yr were converted to other crops, thereby males handled all food types more slowly than precluding their availability as roost sites. males (F,, 3 = 5.169, P = 0.019; Table 3). In Birds began arriving at nocturnal roosts about addition, the type of food item had a significant 1.5 hr before sunset and began departing IO-30 effect on processing time (F,, 253= 47.552, P < min after sunrise. In late March, as the migration 0.001). Because its seed is encased in a hard period approached, birds departed earlier (some- capsule, Rottoboellia cochinchinensis took lon- times before sunrise) and returned later. ger to process than other seeds. There was no It took nearly 1 hr for large roosts to fill in significant interaction between sex and food the evening, but in the morning they usually item, implying that the effect of food items on emptied in less than 30 min. Departures were handling time was the same in both sexes. spectacular as hundreds of thousands of birds When we observed entire feeding bouts, the left together in a burst of noise and wind. Some- longest bouts involved immature green sorghum, times they spiraled upward, forming tomado- where a female ate 52 seeds in 241 set and a shaped funnels, and at other times broad col- male ate 80 seeds in just under 290 sec. Because umns left in serpentine fashion, with the lead birds were less conspicuous when feeding on birds above those behind. Small roosts (< rice and wild grasses, duration of feeding bouts 100,000 birds) sometimes emptied in one flight, on those items was more difficult to document. but in large roosts wave after wave emerged un- The longest bouts in rice (17 seeds in 54 set) til only a few stragglers remained. Departures and Orizya latifolia (15 seeds in 40 set) were usually took place from one part of the roost, both for males. but in the largest roosts we observed up to three distinct routes and directions of departure. ROOSTING We observed birds roosting on the ground, Dickcissels not only fed in large flocks but both in sugarcane roosts as well as other roosts. roosted together at night in very large aggrega- On several occasions, large numbers crowded tions. In Portuguesa and Cojedes they roosted at together on furrows in sugarcane roosts. We also night almost exclusively in sugarcane, a tall, observed individual Dickcissels entering the sturdy grass with numerous horizontal leaves roost after dark, their presence indicated by a that provide ideal perches and cover from pred- distinctive buzzerlike flight call. Although ators. When sugarcane was not available, as in ffrench (1967) found that small roosts were of- Gu&ico, the birds roosted in dry cattails (Typha ten quiet at night, at large roosts we could hear sp.), grass, shrubs, small acacia (Acacia) trees, Dickcissels throughout the night, fleeing from and a fallow rice field. predators such as Barn Owls (Tyto alba). In ad- Birds returning to roost in sugarcane did so in dition, we observed small groups (< 100 birds) an orderly fashion; they wasted little flight time returning to the roost at dawn prior to the de- or effort before entering the roost. Birds return parture of any individuals. We presumed these ing to non-sugarcane roosts, however, appeared birds were unable to reach the roost the previous more reluctant to settle down and were often ob- evening and were returning at first light to rejoin served flying in several directions before finally the larger flocks. entering the roost. Large nocturnal roosts sometimes subdivided The same nocturnal roosts were often used into dozens of daytime roosts which varied in every year. Of the 19 roosts located in our study size from a few tens of individuals to hundreds DICKCISSELS IN WINTER-Basili and Temple 295 of thousands. After foraging for a few hours, were mounted on birds at the same roost, and at Dickcissels entered daytime roosts close to their that time the next closest roost was 10 km away. food source and remained there until late-after- Birds with transmitters were most easily detect- noon feeding bouts. When birds were inactive, ed at densely populated nocturnal roosts. They we could locate daytime roosts by their loud were sometimes located in daytime roosts, but chatter (when birds were alarmed, chatter im- this proved more difficult as birds were widely mediately stopped, then resumed shortly later). distributed on the landscape. Shortly after we Daytime roosts usually included isolated trees, began the telemetry work, the roost where we forest stands, narrow forest strips along canals fitted birds with transmitters began to get smaller and streams, hedgerows, grasses, and sugarcane. and a new roost formed 5 km closer to the main Only rarely was sugarcane used as a daytime feeding areas. An additional roost was discov- roost, but on several occasions we observed ered nearby, bringing the total to four nocturnal groups of fewer than 100 birds return to their roosts within 100 km2. nocturnal sugarcane roost after early morning Individual birds were not strictly faithful to a feeding. Selection of daytime roosts was likely single nocturnal roost. All 14 birds with trans- determined by proximity to feeding area and mitters used multiple nocturnal roosts; eight in- shelter from the sun. dividuals used two roosts, five individuals used In late March and, April, Dickcissels often three roosts, and one individual used four roosts. shared sugarcane roosts with other nearctic-neo- Of the eight birds that used two roosts, five tropical migrants. Bobolinks (Dolichonyx oryzi- switched roosts once and remained at the new vorus), migrating north from Brazil, formed location, whereas three individuals moved back mixed-species roosting and foraging flocks with and forth between the two roosts. The exchange Dickcissels in the Venezuelan llanos. Bobolink of birds between neighboring roosts suggests flocks varied in size from a few to 5,000 indi- that individual roosts do not represent distinct viduals. Premigration flocks of thousands of subpopulations. Barn Swallows (Hirundo rustica) and Bank Nocturnal telemetry observations confirmed Swallows (Riparia riparia) were also observed that Dickcissels made nighttime flights and used roosting with Dickcissels, although the swallows several feeding areas. One individual left a roost arrived later and departed earlier than the Dick- at 2200 and flew to a neighboring roost 10 km cissels. away, where it was detected at 0200. Birds with transmitters did not always return to the same ROOST SIZE daytime roosts, indicating flexibility in daily We counted Dickcissels in 16 roosts that choice of feeding areas. ranged in size from 20,000 to 2,950,OOO birds. The median count was 580,000, and nearly 40% CAUSESOF MORTALITY of all roosts were larger than 1 million birds. In Overwinter mortality is an important factor in Trinidad, ffrench (1967) reported 66,000 2 the dynamics of the Dickcissel population. We 15,000 birds in the Oropuche Lagoon roost in observed 17 species (14 avian, 3 mammalian) 1962. preying on Dickcissels in Venezuela: Merlin Because most Dickcissels overwinter in Ven- (Falco columbarius), Aplomado Falcon (F. fe- ezuela, we think roost counts provide a unique moralis), Peregrine Falcon (F. peregrinus), Bat opportunity to estimate the total Dickcissel pop- Falcon (F. rufigularis), American Kestrel (F. ulation. Between 24 January and 20 February sparverius), Pearl Kite (Gampsonyx swansonii), 1993, we located and estimated the sizes of all White-tailed Kite (Elanus caeruleus), Roadside roosts in Portuguesa, Cojedes, and Guarico. We Hawk (Buteo magnirostris), Short-tailed Hawk discovered five roosts with a total Dickcissel (B. brachyurus), Gray Hawk (B. nitidus), Harris ’ population estimated at more than 6 million Hawk (Parabuteo unicinctus), Long-winged birds. Four roosts in the Acarigua area were es- Harrier (Circus bufSoni), Barn Owl, Short-eared timated to hold 2,370,OOO; 1,945,OOO; 590,000; Owl (Asio Jlammeus), jaguamndi (Felis yagoua- and 420,000 individuals, respectively. One roost roundi), grisonlhuron (Galictis vittata), and tay- near Calabozo contained 1,125,OOO individuals. ra (Eira barbara). On Trinidad, ffrench (1967) Since these estimates represent a majority of the noted three predators, all avian, of Dickcissels. entire Dickcissel population, approximately 30% Merlins, also migrants, were the most com- of the entire population can be found roosting mon diurnal predators in Venezuela. We gener- together in a single sugarcane field. ally found one to five Merlins at every roost. Barn Owls were the most conspicuous nocturnal RADIO TELEMETRY predators. Although Peregrine Falcons were not In January and February 1993 we monitored common in the llanos, they appeared at Dickcis- 14 individuals with radio transmitters. All radios se1 roosts in March and April, presumably on 296 STUDIES IN AVIAN BIOLOGY NO. 19 their migration north. On 15 April 1995, 19 Per- A few birds still disperse to historical parts of egrine Falcons hunted at a large Dickcissel roost their winter range (e.g., appearing every few near the village of El Cruce southeast of Aca- years in Trinidad), but most now congregate in rigua. Jaguarundis were the most common mam- agricultural regions of central Venezuela. malian predators we observed. Our research suggests that most Dickcissels overwinter in the vicinity of Acarigua. This may CONFLICTS BETWEEN DICKCISSELS AND represent a population shift since Fretwell VENEZUELANFARMERS (1977) indicated that most birds wintered near The greatest cause of overwinter mortality in Calabozo, approximately 220 km east of Acari- Venezuela appears to be deliberate chemical poi- gua. Our research suggests that the agricultural soning by farmers, who consider Dickcissels areas around Acarigua and Calabozo remain the pests. We observed numerous birds that showed center of abundance from November through signs of pesticide poisoning (e.g., loss of bal- April. We did not find a December movement ance, respiratory problems, paralysis), and on of Dickcissels out of Venezuela (cf. Fretwell one occasion we discovered a roost that 3 d ear- 1977). lier had been fumigated with a cannon sprayer “Short-stopping” (Hestbeck et al. 1991) of attached to a tractor. Approximately 1,000 Dick- Dickcissels may also now be occurring on the cissels and unknown numbers of migratory species ’ southward migration through Central Bank and Barn swallows were killed. We con- America, as evidenced by large roosting flocks firmed that at least five other nocturnal roosts in Central American agricultural areas. Early in had been chemically targeted since 1989 (G. Ba- their fall migration, Dickcissels have been ob- sili, pers. comm. with farmers). One farmer ac- served flocking with blackbirds (Icteridae) and knowledged that every few years he had sprayed feeding on rice in the Arkansas Grand Prairie a roost on his property with an aerial application (Meanly 1971). Although few Dickcissels over- of parathion (spraying is done in early morning winter in the United States, a large proportion prior to the birds ’ departure to feeding areas). of those that do can be found in the rice lands The chemical is so effective that most roosting of the alluvial plain of the Mississippi River and birds are killed. The farmer described dead coastal prairies of Texas and Louisiana (Root Dickcissels as “knee-deep.” He estimated that 1988). This suggests that some Dickcissels may control efforts on his property have killed more be short-stopped before leaving the United than 1 million birds over the years, and neigh- States. Dickcissels appear to be able to adapt to boring farmers and agricultural workers corrob- rapidly changing agricultural landscapes along orated his statements. their migration route (e.g., Slud 1964). Thus, we think the winter range of Dickcissels will vary DISCUSSION as agricultural practices change in Venezuela WINTER DISTRIBUTION and elsewhere in northern South America, as The present concentration of Dickcissels in well as in the countries through which Dickcis- the states of Portuguesa and Gutico, Venezuela, sels migrate. is best explained when one reviews the condi- tion of the llanos before mechanized agriculture ROOSTING changed the landscape. We believe Dickcissels Nocturnal roosts in Venezuela were much historically foraged on the seeds of grasses and larger than roosts reported from Trinidad forbs, resources that were more uniformly dis- (ffrench 1967). Sugarcane, when available, ap- tributed on the landscape (Levey and Stiles peared to be the preferred habitat for night roost- 1992). After entering the llanos of northern ing. Sugarcane was introduced to Venezuela in South America, Dickcissels probably dispersed 1520 but was not intensively cultivated until the widely in search of productive feeding areas. 1920s (Gomez-Alvarez 1975). Dickcissels have Museum specimens support this idea and indi- quickly selected sugarcane as preferred roosting cate that until the first half of the twentieth cen- habitat. When sugarcane was not available, tury Dickcissels occurred throughout the llanos Dickcissels roosted in bamboo, cattail marshes, of northern South America (Venezuela, Colom- grasses, and shrubs (ffrench 1967; Orians and bia, Brazil, Guyana, and Trinidad), including sa- Paulson 1969; G. Basili, pers. obs.). We suspect vannas of the Amazon basin. Since the 195Os, that these latter habitats are similar to preagri- the llanos have become an agricultural region culture roost sites. dominated by rice and sorghum. Seeds are no longer distributed uniformly but are superabun- DICKCISSELMORTALITY dant and clumped. Instead of dispersing over a Humans kill Dickcissels, sometimes acciden- wide area, most Dickcissels remain concentrat- tally but most often intentionally. Accidental ed, taking advantage of the concentrated food. Dickcissel mortality occurred during our study DICKCISSELS IN WINTER-Bnsili artd Temple 297 when sugarcane roosts were harvested while Venezuelan farmers (J. L. Mendez-Arrocha, they were still occupied by Dickcissels. Prior to pers. comm.), and lethal control of Dickcissels harvest, fields were burned to eliminate excess, was most flagrant (G. Basili, pers. comm. with nonvaluable leaf material. Roosts burned at farmers). We think lethal control may have been night have been known to kill some birds responsible for the rapid Dickcissel decline ob- (ffrench 1967; G. Basili, pers. comm. with farm served in North America between 1966 and ers 1991-1994). We only observed one roost be- 1978. ing burned; it was in the late afternoon when By the early 198Os, crop production in Ven- birds were settling in for the evening, and no ezuela had increased dramatically, yet Dickcis- mortality was discovered. The impact of roost- se1 populations in North America were reduced. burning while birds are present remains unclear The impact of Dickcissels on regionwide crop but warrants further investigation. yields probably became less important, but on People eat Dickcissels throughout their range local scales the threat of Dickcissel depredation in Venezuela. Children hunt them with sling- remained high and some farmers still suffered shots and by throwing short stalks of sugarcane substantial economic hardships (Basili and Tem- into dense flocks arriving at nocturnal roosts. ple 1998). Because of this history, lethal control Some people enter roosts at night and club the persists in Venezuela, and it continues to play a birds with sticks and bats. When shotguns are critical role in Dickcissel population dynamics. used to protect crops from Dickcissel depreda- If our ideas concerning the dynamics of the tion, birds that are shot are eaten. The most un- Dickcissel population are correct, Dickcissels usual hunting method we observed was a vehi- may be in jeopardy. Because the population has cle driving rapidly through flocks flying low declined rapidly, and because single roosting ag- across farm roads. It seems unlikely that this gregations sometimes comprise approximately cause of mortality has a major effect on the 30% of the species ’ entire population, Dickcis- Dickcissel population, yet these daily events sels continue to be vulnerable to catastrophic may be important when summed over the spe- mortality. Lethal control of Dickcissels is still cies ’ 7-mo stay in the llanos. practiced (Basili and Temple 1998). The fact that Bobolinks, Barn Swallows, and There appears to be a major conservation Bank Swallows occupied Dickcissel roosts in problem regarding the Dickcissels’ central win- March and April is an important conservation tering area in Venezuela, and it is reasonable to issue because these species now share the same infer that the entire population may be limited risk of intentional chemical poisoning. This be- as a result. These results support Sherry and came apparent when we discovered dead Barn Holmess’ (1995:86) statement that “understand- and Bank swallows among dead Dickcissels fol- ing the whole migratory phenomenon is essen- lowing an April control operation. Because Bob- tial for effective conservation and manage- olinks feed with Dickcissels in rice, Bobolinks ment.” Once reasonable inferences can be made are also susceptible to control operations in about population limitations, that knowledge feeding areas. These observations, and knowl- should be applied to conservation action. In the edge of rice and sorghum depredations on the case of the Dickcissel, we recommend that con- Bobolinks’ main wintering grounds in Brazil and servation efforts be directed primarily toward re- Argentina (Sick 1986), raise the question of ducing lethal control in the Venezuelan llanos whether Bobolinks are subject to similar threats while also addressing the concerns of local of chemical poisoning as Dickcissels. farmers. If the predicament of Dickcissels wintering in Venezuela is ignored, a critical oppor- FARMERS AND DICKCISSEL POPULATION TRENDS tunity for effective and efficient management Most farmers use nonlethal controls to keep and conservation of Dickcissels, and other mi- birds out of their fields, but some farmers use gratory birds, will be lost. toxic agricultural chemicals to kill Dickcissels (Basili and Temple 1998). We think that events ACKNOWLEDGMENTS on the wintering grounds in Venezuela could This study was funded by the Wildlife Conservation have been responsible for most of the 40% Dick- Society, National Fish and Wildlife Foundation, Lin- cissel population decline reported in North coln Park Zoo Scott Neotropic Fund, Zoological So- America since the late 1960s (Peterjohn et al. ciety of Milwaukee County, and Agricultural Experi- 1995). When the decline was most precipitous ment Station and Department of Wildlife Ecology at the University of Wisconsin-Madison. Field assistance (1966-1978), crop production was still limited. was provided by A. Basili, I? Desenne, C. Guglielmo, At this point there were many Dickcissels but .I. Hansen, R. Israel, M. Lentino, C. Rodner, C. San- few crops on the landscape, possibly resulting in chez, and T. Teal. We thank G. Aymard of Unellez- severe crop depredation. During this period, Guanare for help in botanical identification, S. Mc- Dickcissels were considered a major problem fox Williams for help in interpreting diet information, and 29% STUDIES IN AVIAN BIOLOGY NO. 19

J. Cary for assisting with data analysis. The Sociedad Finch (editors). Ecology and management of neotro- Conservacionista Audubon de Venezuela, EcoNatura, pica1 birds: a synthesis and review of critical issues. G. Morales (Instituto de Zoologia Tropical, Universi- Oxford University Press, Oxford, U.K. dad Central de Venezuela-Caracas), E. Lander (Uni- MAURER, B. A., AND M. A. VILLARD. 1996. Continental versidad Central de Venezuela-Maracay), D. Aguerro scale ecology and neotropical migratory birds: how (FONIAP), and A. Grajal (Wildlife Conservation So- to detect declines amid the noise. Ecology 77: l-2. ciety) provided critical logistical support. We thank F? MEANLEY, B. 1971. Blackbirds and the southern rice D. Vickery and J. R. Herkert for reviewing and im- crop. U.S. Fish and Wildlife Service Resources Pub- proving the manuscript. We thank P Bichier, l? Desen- lication 100. ne, and S. Strahl for their support and encouragement. MEYER DE SCHAUENSEE,R., AND W. H. PHELPS, JR. Finally, we thank the people of the llanos for their 1978. A guide to the birds of Venezuela. Princeton hospitality and for opening their doors for hot meals, University Press, Princeton, NJ. cold drinks, showers, and wonderful company. MONASTERIO, M. 1970. Ecologia de las sabanas de AmCrica tropical. II. Caracterizacidn ecblogica de1 LITERATURE CITED clima en 10s llanos de Calabozo, Venezuela. Revista Geografica 21:5-38. ANUARIO ES~ADISTICOAGROPECUARIO. 1996. Repdblica NEWTON, I. 1972. Finches. Taplinger Publishing, New de Venezuela. Ministerio de Agricultura y Cria, Di- York, NY. recci6n General Sectorial de Planificaci6n de1 Sector ORIANS, G. H., AND D. R. PAULSON. 1969. Notes on Agricola. Direcci6n de Estadistica. Caracas, Vene- Costa Rican birds. Condor 71:426-431. zuela. BASILI, G. D., AND S. A. TEMPLE. 1998. Dickcissels PETERJOHN,B. G., J. R. SAUER, AND C. S. ROBBINS. and crop damage in Venezuela: defining the problem 1995. Population trends from the North American with ecological models. Ecological Applications 9: Breeding Bird Survey. Pp. 3-39 in T. E. Martin and 132-739. D. M. Finch (editors). Ecology and management of FERNANDEZ-YEPEZ,A. J. 1945. El Problema de1 Pajaro neotropical birds: a synthesis and review of critical Arrocero. Ph.D. dissertation. Universidad Central de issues. Oxford University Press, Oxford, U.K. Venezuela, Caracas, Venezuela. PETIT, D. R., J. E LYNCH, R. L. HUTTO, J. G. BLAKE, FFRENCH,R. P 1967. The Dickcissel on its wintering AND R. B. WAIDE. 1995. Habitat use and conserva- grounds in Trinidad. Living Bird 6:123-140. tion in the neotropics. Pp. 145-200 in T. E. Martin FRETWELL, S. D. 1977. Is the Dickcissel a threatened and D. M. Finch (editors). Ecology and management species? American Birds 3 1:923-932. of neotropical birds: a synthesis and review of crit- FRETWELL, S. D. 1986. Distribution and abundance of ical issues. Oxford University Press, Oxford, U.K. the Dickcissel. Pp. 211-239 in R. E Johnston (edi- PHELPS,W. H., AND W. H. PHELPS,JR. 1963. Lista de tor). Current ornithology, vol. 4. Plenum Press, New las Aves de Venezuela su distribucibn. 2d ed. Vol. York, NY. 1, pt. 2. Boletin de la Sociedad Venezolana Ciencias FRETWELL, S. D., AND T. G. SHANE. 1975. Ecotypic Naturales 24, no. 104 and 105. variation in wintering Dickcissels. Eastern Bird PYLE, F?, S. N. G. HOWELL, R. P YUNICK, AND D. E Banding Association News 38:125-128. DESANTE. 1987. Identification guide to North Amer- GOMEZ-ALVAREZ, E 1975. CaAa de Azucar. Fondo Na- ican passerines. Slate Creek Press, Bolinas, CA. cional de Investigaciones Agropecuarias. Caracas, RAPPOLE,J. H., AND A. R. TIPTON. 1991. New harness Venezuela. design for attachment of radio transmitters to small GROSS, A. 0. 1921. The Dickcissel (Spiza americana) passerines. Journal of Field Ornithology 62:335- of the Illinois prairies. Auk 38:163-184. 337. HESTBECK,J. B., J. D. NICHOLS, AND R. A. MALECKI. ROOT, T L. 1988. Atlas of wintering North American 199 1. Estimates of movement and site fidelity using birds. University of Chicago Press, Chicago, IL. mark-resight data of wintering geese. Ecology 72: SARMIENTO, G. 1984. The ecology of neotropical sa- 523-533. vannas. Harvard University Press, Cambridge, MA. HILTY, S. L., AND W. L. BROWN. 1986. A guide to the SAUER, J. R., B. G. PETERJOHN,S. SCHWARTZ, AND J. birds of Colombia. Princeton University Press, E. HINES. 1996. The North American Breeding Bird Princeton, NJ. Survey home page. Ver. 95.1: www.mbr.nbs.gov/ LEVEY, D. J., AND E G. STILES. 1992. Evolutionary bbs/bbs.html. Patuxent Wildlife Research Center, precursors of long distance migration: resource Laurel, MD. availability and movement patterns in neotropical SHERRY, T. W., AND R. T. HOLMES. 1995. Summer ver- landbirds. American Naturalist 140:447-476. sus winter limitation of populations: what are the LINDSTR~M, A. 1989. Finch flock size and risk of hawk issues and what is the evidence? Pp. 85-120 in T. predation at a migratory stopover site. Auk 106: E. Martin and D. M. Finch (editors). Ecology and 225-232. management of neotropical birds: a synthesis and LITVAITIS, J. A., K. TITUS, AND E. M. ANDERSON. 1994. review of critical issues. Oxford University Press, Measuring vertebrate use of terrestrial habitats and Oxford, U.K. foods. Pp. 254-274 in T A. Bookout (editor). Re- SICK, H. 1986. Ornitologia Brasileira, Uma Introdu@o. search and management techniques for wildlife and 2d ed. Vol. 2. Editora Universidade de Bra&a, Bra- habitats. 5th ed. Wildlife Society, Bethesda, MD. sflia, D.E, Brazil. MARTIN, T E., AND D. M. FINCH. 1995. Importance of SLUD, P. 1964. The birds of Costa Rica: distribution knowledge and its application in neotropical migra- and ecology. Bulletin of the American Museum of tory birds. Pp. xiii-xvi in T E. Martin and D. M. Natural History 128:1-430. DICKCISSELS IN WINTER-Basili and Temple 299

TEMPLE, S. A. 1995. When and where are shrike Pop- cissel, Spiza americana. Physiological Zoology 38: ulations limited? Proceedings of the Western Foun- 370-389. dation of Vertebrate Zoology 6:6-10. ZIMMERMAN, J. L. 1965b. Carcass analysis of wild and ZIMMERMAN, J. L. 1965a. Bioenergetics of the Dick- thermal stressedDickcissels. Wilson Bulletin 77:55-70.