Protected Cerrado Fragments Grow up and Lose Even Metapopulational Birds in Central São Paulo, Brazil

moderately dense woodland (cerradão) grew in two isolated patches of bushy savanna (cerrado) in central são Paulo over 23 years of bird censuses. various uncommon birds were lost and some forest species were permanently gained. Fall and winter fruits attract long and short-distance migrants. woodpeckers and some birds that nest in their holes seem to disappear during tree growth. some birds in weedy areas nearby disappeared when the pastures replaced these areas, however sugar cane reduced the numbers of birds even more up to the point when some areas became pastures once more. even travel-prone species disappear with vegetation growth in cerrado protected fragments, and therefore " metapopulations " may not survive over time, only in space. RESUMO Fragmentos protegidos de cerrado crescem e perdem mesmo aves metapopulacionais na região central do Estado de São Paulo, Brasil


INTRODUCTION
as agriculture develops, cerrado (savanna) areas in central Brazil are disappearing.only some fragments are protected, usually small ones without corridors, although some large parks have been set aside. in 1982, i started studying birds in open or "campo cerrado" zones near itirapina in central são Paulo state (willis, 2004).since then, i have also observed birds from some semi-open "cerrado sensu strictu" areas, notably two patches east of itirapina, one being an area protected by a university (UnesP) and the other by cane planters.
these two patches are fragments of a large cerrado zone that was mostly destroyed in 1981 for pastures and sugar cane (r.monteiro, pers.comm.) on both sides of the main highway northwest between the cities of rio claro and são carlos.although the wooded savanna was low or irregular at the time, these areas have grown so much that they have become cerradão woodland.
common secondary forest birds are lost in isolated zones as tall forests grow back (willis, 1974), and here it was found that similar losses of uncommon cerrado birds occur in isolated cerrados with regrowth.some woodland species have moved in, but most that wander in have been unable to maintain metapopulations.nearby open zones also lost other birds when they were widely planted with relatively birdless sugar cane.

STUDY AREAS AND METHODS
the ´UnesP cerrado` is a 38.7-ha fragment at 865 m and 22° 14' s and 47° 41' w, which has been surrounded by pastures and sugar cane since 1981.this area is on top of a flat and sandy plateau or tableland, which drops off to formerly forested lowlands just east and south.wooded zones and creeks occur irregularly along the drop-off, and some have been protected.scattered woodlots are protected in the lowlands, for instance those of Fazenda são José near rio claro (willis & oniki, 2002).scattered cerrado zones occur off west, but there are mostly pastures and cane, orange or tree plantations.
two km west, across cane fields and beyond km 200 on the washington Luiz main highway (at 850 m, and 22° 15' s and 47° 43' w, on the border of the townships of corumbataí to the east and itirapina to the west), the similar-sized but scrubbier ´km 200` cerrado fragment was observed occasionally.surrounded by roadsides and cane, it is also near narrow wooded drop-offs from the edge of the ´chapada`.some areas were burnt in 1982 and some trees were cut.as in the case of the cerrado at UnesP, many areas have grown into shady cerradão since 1981.
transect censuses listed numbers of birds heard or seen in 74 visits (260.7 h) at UnesP and 14 visits (35.7 h) at the km 200 areas, 8 april, 1982200 areas, 8 april, to 1 February, 2005. .Birds of surrounding open (or pond) areas were noted separately, as were birds flying over.Here, birds are noted "per 100 h."Furthermore, some UnesP visits were with students.
Use of these areas by Picazuro Pigeons (Columba picazuro) invading the state was registered by willis & oniki (1987).thrushes after army ant swarms were noted by willis (1984) and the presence of one species of Hylophilus in willis (1991). willis & oniki (2003) briefly noted birds up to 1998 with dates and times of censuses.

RESULTS
open-area birds -various scrubby or grazed pastures around UnesP were planted with sugar cane in 1984, and then they became open pastures in some areas after 2000.cane plantations and more efficient cattle use led to lower numbers of some birds from 1984 onwards (table 1).Birds that disappeared or almost disappeared included three hawks, one anumbius (which also disappeared in most other areas regionally, perhaps due to recent warm and dry years; willis, 2004), plus two Synallaxis and a Thamnophilus of weedybushy zones.the loss of Gnorimopsar could be due to catching cage birds, as in the region in general (willis, 2004).Geothlypis, anthus, Stelgidopteryx and Xolmis velata continued to be less common even after several cane areas returned to pastures.some others were never very common (Nothura, Falco spp., Tyto, Bucco, Xolmis cinerea, Notiochelidon, Progne, Molothrus, Sporophila spp.).others became more common again when pastures replaced several cane areas (Syrigma, Caracara, Vanellus, athene, Guira, Zenaida, Crotophaga, Tachycineta, Mimus).some regular open-zone birds changed little in abundance despite the cane plantations: Crypturellus, Milvago, Herpetotheres, Cariama, Colaptes, Furnarius.Bubulcus (and woodlotnesting and roosting Columba picazuro) have moved into the region in recent years.Sicalis citrina and Pseudoleistes have appeared recently in small numbers.Sicalis luteiventris flocks have been reported by local pasture owners (it is increasing regionally where ungrazed Brachiaria is allowed to seed; antunes & willis, 2003).a recent rarity of Volatinia may be due to the loss of woodlot grass, rather than to pasture or cane effects; it is still common in the locally grassy km 200 site.
water birds were little studied as they are mostly at rather distant ponds or creeks downhill; water areas and creeks are often lacking in protected areas in the state.a small gully that crosses the km 200 area has some water after rain and attracts seed-eating birds., d -1982-83; B -1984-97; c, e -2000-05; F -Flying over; and w -water birds.(remsen, 2001).occasional winter birds from the south (Leucochloris, Phibalura, Contopus, Turdus albicollis, Platycichla) use cerradão zones with flowers or fruit, while passage migrants from the south (elaenia parvirostris, e. mesoleuca, Turdus nigriceps, Tityra) also use fruit.the cerradão seems to have more fruit than cerrado here, and therefore most of these migrants probably still occur.the local migrant elaenia obscura is abundant in fall and winter, using local Miconia, Pera glabrata and other fruits that are even more common now; it nests only in gallery woods near creeks in the region.two semi-resident thrushes, (Turdus leucomelas and T. amaurochalinus) also enter commonly in the fall and winter at the same time as small trees produce fruit in the cerradão.
the former species is now more common, as elsewhere in the interior of the state, after having recent dry and warm years (perhaps replacing T. amaurochalinus somewhat).Both species occasionally nest in the cerradão fragments here, and followed Labidus ants in the UnesP area years ago (willis, 1984).recently only the km 200 area has had swarms and therefore army ants may have become extinct in the UnesP fragment (eciton burchelli extinct in both areas).
Tersina, Myiozetetes and Pitangus ("local migrants") also wander in for fruit in fall and winter, without nesting.Myiozetetes has become more common and Pitangus less so as the cerradão has grown over the years.ramphastos toco also visits for fruit, though it needs large trees to nest, and is a species that is increasing in the region (willis & oniki, 2002; willis, 2003).increases in dacnis and Saltator wanderers in recent years could also be due to fruit or growth of trees.some fruit trees could have been introduced by fruit-eating birds that wander in (m.Pinheiro, pers. comm.).some other species that are now more common are understory fruit eaters (Chiroxiphia, antilophia) and they occur locally mostly in gallery or escarpment woods.Myiarchus ferox also seems to wander in from nearby gallery woods more, perhaps as it uses fruit and prefers woodlands and not hot cerrados.M. tyrannulus has decreased, even though it uses fruit, perhaps as it uses more open cerrados and edges or has lost cavities for nests.the loss of summer M. swainsoni lately is unexplained, or due to the loss of woodpecker holes (below) to nest, as it uses both the woodland canopy and edge, even woods too tall for ferox.
other species of the sixteen that have increased in number do not always use fruit, but prefer shady woodlands and avoid cerrados: Piaya, Lathrotriccus, Tolmomyias, summer Vireo, Basileuterus hypoleucus.Columba picazuro has become more common in the state in recent years (willis & oniki, 1987), perhaps causing some decrease in C. cayennensis here due to use of perches to sing or nest (the latter eats fruits, the former seeds more in open zones; they perhaps compete somewhat for food).Tolmomyias males try to establish midlevel or understory metapopulations even on the rio claro campus or in a suburb in Barão geraldo (campinas), but here and there do not seem to succeed; stable populations and nests are in somewhat taller woods, even a small woodlot at sítio novo Horizonte near rio claro (pers.obs.). it has not been in the UnesP cerrado except for July, 2000to January, 2003, and km 200 in June, 1982and February, 2005. some

DISCUSSION
even though cerrado and open-area birds can generally fly to distant habitat fragments, they (and nearby "travel-prone" or metapopulational woodland birds) are not always able to establish new populations or keep old ones going.a major problem is that protected fragments tend to grow, or are only rarely pruned or catch fire.thus, the "metapopulation" theory which was very prominent in the literature some years ago is mostly applicable to open habitats in arctic or sub arctic regions, where vegetation does not grow much even if preserved.the theory applies to patchiness of habitat in space, not over time.
in this case, some species increased and others remained in growing bushy savannas, now medium-tall cerradão.Unfortunately, many woodland species that came in from outside tended to be rare or not establish local reproduction, while several bushy-cerrado birds declined to local extinction.some of these latter species are common, but a few are rather rare and subject to regional extinction (elaenia cristata, Neopelma, Casiornis, euscarthmus).
it may be that, in the corumbataí/itirapina region, some cerrado and campo zones will not grow because of sandy soil.However, even open sandy savannas protected from fire near itirapina are growing into dense herbaceous vegetation or bushy zones, rather than staying open (willis, 2004).
one cannot assume that birds able to cross agricultural or pasture zones between habitat fragments will be able to find habitat and mates and reproduce enough to reestablish populations.the "metapopulation" idea was quite optimistic about this possibility, but was mainly espoused by workers in temperate/arctic zones where vegetation is naturally low and grows slowly (and birds are mostly migrants which fly easily from one patch to another).northern scientists often thought linking patches with corridors would not be necessary and that birds would quickly repopulate distant areas after loss or habitat changes.a lack of connecting low or medium-tall corridors may be a problem in the future, as well as a lack of fire and other management methods.Larger areas of cerrado, off the north and west, may preserve the species lost here, if current enthusiasm about using the savannas for cattle, cane, orange, and soybean production eventually wanes.cane, as shown here, has few birds and causes losses of even some pasture birds.
acknowledgments -i would like to thank the Universidade estadual Paulista, the Brazilian research council (cnPq) and the research Foundation of são Paulo (FaPesP).Publication no.47 of the institute for studies of nature.

TABLE 1 Continued...
16 species have maintained populations in woodland growth: Penelope and Leptotila eating fruit (and border elaenia flavogaster, Tyrannus melancholicus, Tangara and Thraupis sayaca fruit and insects); Nyctidromus and Hydropsalis night insects at the edges (though the latter needs semi open edges to nest and spend the day); Five that eat herb seeds or insects in semi-open cerrado or on the edges (Columbina spp, Schistochlamys, Coryphospingus, Zonotrichia) have become rarer and C. squammata and Schistochlamys are extinct in the area.Tapera, a locally extirpated edge species that mostly parasitizes nests of Synallaxis, lost two open-area hosts but not wood-edge S. frontalis; perhaps the latter hides nests in the cerradão.two edge hummingbirds (Phaethornis, Hylocharis) and woodpeckers (Colaptes melanochloros, Veniliornis) are perhaps not adapted for very shady cerradão.growing trees perhaps rarely provide dead trunks to excavate nests.there was a complete loss of insectivorous Casiornis, which nests in woodpecker holes and uses moderately open cerradão, paralleling losses of related Myiarchus spp. it has disappeared recently in other cerradões near itirapina (studies in progress).melletti & Penteriani (2003) found few woodpeckers in regrowing woods and Fort & otter (2004) found less success of a hole-nesting bird in a regrowing forest in canada.small insectivorous flycatchers of cerrado are also disappearing here and in other growing cerradões near itirapina: euscarthmus, Hemitriccus margaritaceiventer, Phaeomyias (mostly a summer bird, working low canopies), plus a greenlet Hylophilus.Cyclarhis and Parula of open canopies have decreased, but are still fairly common.an understory bird of cerrado or gallery edges, now absent in the whole region, was only seen in november 1983 (Neopelma).other campo and cerrado birds of the region (see willis, 2004) were not recorded, or only in or near the km 200 more open habitat (Lepidocolaptes; Formicivora, now gone there; Suiriri suiriri and Melanopareia torquata and Saltator atricollis were recorded only just west in other patches of more open cerrado).the UnesP and km 200 areas were probably too dense for the last three species, even in 1982.