Plant Sexual Systems and a Review of the Breeding System Studies in the Caatinga, a Brazilian Tropical Dry Forest

Abstract

• Backgrounds and Aims The reproductive biology of a community can provide answers to questions related to the maintenance of the intraspecific pollen flow and reproductive success of populations, sharing and competition for pollinators and also questions on conservation of natural habitats affected by fragmentation processes. This work presents, for the first time, data on the occurrence and frequency of plant sexual systems for Caatinga communities, and a review of the breeding system studies of Caatinga species.

• Methods The sexual systems of 147 species from 34 families and 91 genera occurring in three Caatinga areas in north-eastern Brazil were analysed and compared with worldwide studies focusing on reproductive biology of different tropical communities.

• Key Results The frequency of hermaphrodite species was 83·0 % (122 species), seven of these (or 4·8 % of the total) being heterostylous. Monoecy occurred in 9·5 % (14) of the species, and andromonoecy in 4·8 % (seven). Only 2·7 % (four) of the species were dioecious. A high percentage of hermaphrodite species was expected and has been reported for other tropical ecosystems. With respect to the breeding system studies with species of the Caatinga, the authors' data for 21 species and an additional 18 species studied by others (n = 39) revealed a high percentage (61·5 %) of obligatory self-incompatibility. Agamospermy was not recorded among the Caatinga studied species.

• Conclusions The plant sexual systems in the Caatinga, despite the semi-arid climate, are similar to other tropical dry and wet forest communities, including those with high rainfall levels, except for the much lower percentage of dioecious species. The high frequency of self-incompatible species is similar to that reported for Savanna areas in Brazil, and also for dry (deciduous and semideciduous) and humid tropical forest communities.

Sexual systems, breeding systems, dry forests, Caatinga, north-eastern Brazil

INTRODUCTION

The reproductive biology of tropical species has been studied in order to understand the mechanisms of gene flow and speciation in plant communities (Baker, 1959; Federov, 1966; Bawa, 1974, 1980; Bawa et al., 1985; Bawa and Hadley, 1990; Gibbs, 1990; Ibarra-Manriquez and Oyama, 1992; Bullock, 1995). The types and frequencies of sexual systems, particularly dioecy, have received attention in studies of tropical, temperate and artic floras (Bawa, 1980; Flores and Schemske, 1984; Bullock, 1985; Fox, 1985; Steiner, 1988; Renner and Ricklefs, 1995; Oliveira and Gibbs, 2000). The characteristics of flowers and fruits, and the pollination and sexual systems have also been related to dynamic and successional processes (Ibarra-Manriquez and Oyama, 1992).

Studies on the sexual and breeding systems of tropical forests and their ecological relationships have focused particularly on the Neotropics: Costa Rica (Bawa, 1974; Bawa et al., 1985; Kress and Beach, 1994; Wesselingh et al., 1999); Mexico (Bullock, 1985; Ibarra-Manriquez and Oyama, 1992; Tabla and Bullock, 2002); Panama (Croat, 1979); and Venezuela (Ruiz-Zapata and Arroyo, 1978; Sobrevila and Arroyo, 1982; Enrech et al., 1988; Ramírez and Brito, 1990; Ramírez et al., 1990). In Brazil, studies on this subject are found for Savanna (‘Cerrado’) areas (Saraiva et al., 1996; Oliveira and Gibbs, 2000), coastal dunes (Ormond et al., 1991; Matallana et al., 2005) and Atlantic forest species (Silva et al., 1997). Community studies in other biogeographical areas include the Mediterranean region (Dafni and O'Toole, 1994) and south-east Asia (Kato, 1996; Momose et al., 1998). However, more studies are needed in order to determine whether different sexual patterns occur among the diverse plant communities.

The Caatinga is a semi-arid vegetation, occurring only in Brazil, almost exclusively in the north-eastern region (Sampaio, 1995; Aguiar et al., 2002; MMA, 2002). It is the fourth largest vegetation type in Brazil, after the Amazonian forest, the Cerrado, and the Atlantic forest (Aguiar et al., 2002), covering 734 478 km² (MMA, 2002). It comprises approx. 82 000 km² (82·88 %) of the state of Pernambuco (Hueck, 1972), which in turn is almost 50 % of the north-eastern region and 8·6 % of the country. Recently the Caatinga was recognized as one of ‘Earth's last wild places’, and was classified as one of the 37 ‘Wilderness Areas of the World’ (Gil, 2002). The Caatinga is a tropical dry forest and scrub vegetation, occurring in markedly seasonal areas with low rainfall levels (e.g. 500–750 mm year⁻¹), which are irregularly distributed through the year (3–5 months), with annual temperatures averaging 23–27 °C. Rainfall irregularity causes a hydric deficit during a large part of the year (drought periods) (for details, see Sampaio, 1995; Rodal and Melo, 1999).

Despite its great extent and importance, few studies with ecological information are available, and there is a notable lack of publications focusing on the biology and the dynamics of its species (for a review, see Leal et al., 2003). Studies on the plant sexual systems of Caatinga species are scarce and focus mainly on individual species (Pinheiro et al., 1991; Vogel and Machado, 1991; Lewis and Gibbs, 1999; Locatelli and Machado, 1999; Machado and Sazima, 1995; Kiill and Ranga, 2000a, b, 2003; Quirino and Machado, 2001; Machado et al., 2002; for a review, see also Machado, 1996 and Machado and Lopes, 2002).

The Caatinga areas are suffering increasing anthropic destruction which results in loss of native fauna and flora (Sampaio, 1995; MMA, 2002). As a consequence, the flora and the physiognomy of the vegetation are being markedly altered, and reduced to small fragments, before basic biological or ecological studies are carried out. A broad knowledge of plant reproductive biology may be essential for the maintenance of the biodiversity of fragmented areas in the tropics (Bawa, 1990) and for management projects in this ecosystem.

This work presents general data on the occurrence and frequency of plant sexual systems for Caatinga communities, and includes a review of breeding system studies of Caatinga species. These data complement the study of Machado and Lopes (2004), which reports on floral traits and pollination systems of species of the same Caatinga communities.

MATERIALS AND METHODS

Study sites and species

The study was mainly carried out with species occurring in three semi-arid areas situated in the municipalities of Alagoinha, Buíque and Serra Talhada, located in the rural zone of Pernambuco state, north-eastern Brazil (see fig. 1 in Machado and Lopes, 2004). The areas have different physiognomies, and floristic and phytosociological surveys are available in Ferraz et al. (1998); Rodal et al. (1998) and Figueiredo et al. (2000).

The first area, called Sítio Riacho, is located in the municipality of Alagoinha-PE (8°27′S and 36°46′W), comprises about 80 ha, is at an altitude of 762 m a.s.l. and is 200 km from the coast. The number of dry months varies from five to seven, the mean annual temperature is 22 °C and the precipitation approx. 550 mm (Griz and Machado, 2001). The vegetation is a dense shrubby Caatinga (Egler, 1951).

The second area is a Brazilian National Park located in the Catimbau valley, in the municipality of Buíque-PE (8°67′S and 37°11′W), approx. 285 km from the coast. The altitude ranges from 800 to 1000 m a.s.l., the mean annual precipitation and temperature are, respectively, 1095·9 mm (mostly concentrated between April and June) and 25 °C (SUDENE, 1990). The vegetation is very unusual with plant species not found in other Caatinga areas, some of them typical of ‘Campo Rupestre’ vegetation (Rodal et al., 1998).

The third and most inland area (approx. 700 km from the coast) is Fazenda Saco, an experimental station of the Agricultural Research Company of Pernambuco (IPA), located in the municipality of Serra Talhada-PE (7°59′S and 38°19′W). The altitude of the area is about 600 m a.s.l., the mean annual precipitation is around 650 mm (concentrated between January and May) with a dry period between June and December. The mean annual temperature is 26 °C (Machado et al., 1997). The vegetation is dominated by shrubs 3–4 m tall with a few trees reaching 15 m high (Ferraz et al., 1998). For a more detailed description of each study site see Machado and Lopes (2004).

Sexual systems

The sexuality of the flowers of 147 species was determined directly in the field, from herbarium specimens or from literature. Field trips were made five to six times per year, each one lasting 3–5 d, from January 1994 to September 2002. In most cases, only the morphological expression was considered (without performing breeding system experiments), thus functional dioecy was not considered. The species were classified as hermaphrodite (distinguishing those which were heterostylous), monoecious, andromonoecious or dioecious. Voucher specimens are deposited in the Herbarium of the Universidade Federal de Pernambuco (UFP).

Breeding system studies

With regard to the breeding system studies of Caatinga species, the review provides data for 39 species (the authors' data for 21 species, and an additional 18 species studied by others). Information on natural fruit-set (control), cross-pollination, spontaneous and/or hand self-pollination are included, and these are discussed in relation to other ecosystems. For two species [Angelonia cornigera (Scrophulariaceae) and Waltheria rotundifolia (Sterculiaceae)] manual pollinations were made to supplement information given in previous studies by the authors.

RESULTS AND DISCUSSION

Habit and sexual systems

In total, the habits and sexual systems of 147 species, including various life forms (43·9 % shrubs, 18·9 % trees, 18·2 % herbs, 13·5 % lianas, 4·1 % epiphytes and 1·4 % herbaceous climbers), distributed in 34 families and 91 genera were studied (Table 1). This number represents approx. 16 % of the known Caatinga plant diversity (MMA, 2002; for a review, see Leal et al., 2003), an ecosystem where the shrubby habit predominates (Ferraz et al., 1998; Rodal et al., 1998; Figueiredo et al., 2000).

Table 1.

Habits and sexual systems of the studied species from three Caatinga areas in Pernambuco state, north-eastern Brazil

Families and species*	Habit†	Sexual system‡
ACANTHACEAE (2 genera/4 species)
Harpochilus neesianus Mart.	S	Hm
Ruellia asperula (Nees) Lindau	H	Hm
Ruellia aff. paniculata L.	H	Hm
Ruellia sp. 1	H	Hm
AMARANTHACEAE (1/1)
Gomphrena vaga	H	Hm
ANACARDIACEAE (3/3)
Myracrodruon urundeuva Allemão	T	D§
Schinopsis brasiliensis Engl.	T	D§
Spondias tuberosa Arruda Cam.	T	A
ANNONACEAE (1/1)
Rollinia leptopetala R.E. Fries	T	Hm
APOCYNACEAE (3/4)
Aspidosperma pyrifolium Mart.	T	Hm
Allamanda blancheti A. DC.	L	Hm
Allamanda sp.	S	Hm
Mandevilla tenuifolia (Mikan) Woodson	H	Hm
BIGNONIACEAE (2/2)
Tabebuia impetiginosa (Mart. ex DC.) Standl.	T	Hm
Anemopaegma sp.	L	Hm
BOMBACACEAE (2/2)
Pseudobombax marginatum (A. St.-Hil.) A.Robyns	T	Hm
Ceiba glaziovii (Kuntze) K. Schum.	T	Hm
BORAGINACEAE (1/2)
Cordia globosa (Jacq.) Kunth	S	Ht
C. leucocephala Moric.	S	Ht
BROMELIACEAE (6/8)
Dyckia pernambucana L.B. Sm.	H	Hm
Billbergia porteana Brongn.	E	Hm
Encholirium spectabile Mart. ex Schult.f.	H	Hm
Neoglaziovia variegata (Arr. Cam.) Mez.	H	Hm
Portea leptantha Harms	H/E	Hm
Tillandsia gardneri Lindl.	E	Hm
T. loliacea Mart. ex Schult.	E	Hm
T. streptocarpa Baker	E	Hm
BURSERACEAE (1/1)
Commiphora leptophloeos (Mart.) J.B. Gillett	T	D
CACTACEAE (5/11)
Cereus jamacaru DC.	T	Hm
Harrisia adscendens (Gürke) Britton & Rose	HC	Hm
Melocactus bahiensis (Br.et Rose) Werderm.	H	Hm
M. zehntneri (Britton & Rose) Luetzelburg	H	Hm
Opuntia inamoema K. Schum.	S	Hm
O. palmadora Britton & Rose	S	Hm
Pilosocereus catingicola (Gürke) Byles & G.D. Rowley	T	Hm
P. chrysostele (Vaupel) Byles & G.D. Rowley	T	Hm
P. gounellei (F.A.C.Weber) Byles & G.D. Rowley	S	Hm
P. pachycladus (Werderm.) Byles & G.D. Rowley	T	Hm
P. tuberculatus (Werderm.) Byles & G.D. Rowley	T	Hm
CAPPARACEAE (1/4)
Capparis hastata L.	S	Hm
C. flexuosa (L.) L.	T	Hm
C. jacobinae Moric. ex Eichler	S	Hm
C. yco (Mart.) Eichler	S	Hm
CLUSIACEAE (1/1)
Clusia nemorosa G. Mey.	S	D
COMBRETACEAE (2/4)
Combretum hilarianum D. Dietr.	L	Hm
C. leprosum Mart.	L	Hm
C. pisonioides Taub.	S	Hm
Thiloa glaucocarpa (Mart.) Eichler	S	Hm
CONVOLVULACEAE (4/7)
Evolvulus sp.	L	Hm
Ipomoea acuminata Baker	L	Hm
Ipomoea sp. 1	L	Hm
Ipomoea sp. 2	L	Hm
Jacquemontia densiflora (Meissn.) Hall.	L	Hm
Merremia aegyptia (L.) Urb.	L	Hm
Merremia sp.	L	Hm
EUPHORBIACEAE (8/12)
Acalypha multicaulis Müll. Arg.	S	M
Cnidoscolus urens (L.) Arthur	S	M
Croton argyrophylloides Müll. Arg.	S	M
C. sonderianus Müll. Arg.	S	M
Croton sp.	S	M
Dalechampia sp.	HC	M
Euphorbia comosa Vell.	H	M
Jatropha mollissima (Pohl) Baill.	S	M
J. mutabilis (Pohl) Baill.	S	M
J. ribifolia (Pohl) Baill.	S	M
Manihot cf. pseudoglaziovii Pax & K. Hoffm.	S	M
Maprounea aff. guianensis Aubl.	T	M
KRAMERIACEAE (1/1)
Krameria tomentosa A. St.-Hil.	S	Hm
LAMIACEAE (1/1)
Hyptis martiusii Benth.	H	Hm
LEGUMINOSAE (15/29)
Acacia farnesiana (L.) Willd.	S	Hm
Amburana cearensis (Allemão) A.C. Sm.	T	Hm
Anadenanthera colubrina var. cebil (Griseb.) Altschul	T	Hm
Bauhinia acuruana Moric.	S	Hm
B. cheilantha (Bong.) Vogel ex Steud.	S	Hm
B. pentandra (Bong.) Vogel ex Steud.	S	Hm
Caesalpinia ferrea Mart. ex Tul.	T	Hm
C. pyramidalis Tul.	T	Hm
Calliandra aeschynomenoides Benth	S	Hm
Chamaecrista cytisoides (Collad.) Irwin & Barneby	S	Hm
Chamaecrista ramosa (Vogel) var. mollissima (Benth.) Irwin & Barneby	S	Hm
Chamaecrista sp. 1	S	Hm
Cratylia mollis Mart. ex Benth.	S	Hm
Erythrina velutina Willd.	T	Hm
Lonchocarpus aff. campestris Benth.	T	Hm
Mimosa lewisii Barneby	S	Hm
M. tenuiflora (Willd.) Poir.	T	Hm
Parapiptadenia zehntneri (Harms) M.P. Lima & H.C. Lima	T	Hm
Periandra coccinea (Schrad.) Benth.	L	Hm
Senna acuruensis (Benth.) Irwin & Barneby	S	Hm
S. angulata (Vogel) Irwin & Barneby	T	Hm
S. chrysocarpa (Desv.) H.S. Irwin & Barneby	S	Hm
S. macranthera (Collad.) Irwin & Barneby	S	Hm
S. martiana (Benth.) H.S. Irwin & Barneby	S	Hm
S. rizzini Irwin & Barneby	T	Hm
S. spectabilis (DC.) var. excelsa (Schrad.) Irwin & Barneby	S	Hm
S. splendida (Vogel) Irwin & Barneby	S	Hm
S. trachypus (Benth.) Irwin & Barneby	S	Hm
Zornia sericea Moric.	H	Hm
MALPIGHIACEAE (5/7)
Banisteriopsis schizoptera (A. Juss.) B. Gates	L	Hm
Byrsonima gardneriana Juss.	S	Hm
Byrsonima vacciniaefolia Juss.	S	Hm
Byrsonima sp.	S	Hm
Janusia anisandra (Juss.) Griseb.	L	Hm
Heteropteris sp.	L	Hm
Stigmaphyllum paralias Juss.	S	Hm
MALVACEAE (4/5)
Bakeridesia pickelii Monteiro	S	Hm
Herissantia tiubae (K. Sch.) Briz.	H	Hm
Pavonia humifusa A. St.-Hil.	S	Hm
P. martii Mart. Ex Colla	H	Hm
Sida sp.	H	Hm
ORCHIDACEAE (2/2)
Cyrtopodium intermedium Brade	E	Hm
Stenorrynchus lanceolatus (Aubl.) Rich. ex Spreng.	H	Hm
PASSIFLORACEAE (1/4)
Passiflora foetida L.	L	Hm
P. luetzelburgii Harms	L	Hm
Passiflora sp. 1	L	Hm
Passiflora sp. 2	L	Hm
RHAMNACEAE (1/1)
Ziziphus joazeiro Mart.	T	Hm
RUBIACEAE (2/2)
Coutarea hexandra (Jacq.) K. Schum.	S	Ht
Tocoyena formosa (Cham. & Schltdl.) K. Schum.	S	Hm
SAPINDACEAE (2/2)
Allophylos quercifolius (Mart.) Radlk.	S	M
Serjania comata Radlk.	L	M
SAPOTACEAE (1/1)
Bumelia sartorum Mart.	T	Hm
SCROPHULARIACEAE (3/7)
Angelonia bisaccata Benth.	S	Hm
A. cornigera Hook.	H	Hm
A. hirta Cham.	H	Hm
A. hookeriana Gardn.	S	Hm
A. pubescens Benth.	H	Hm
Bacopa sp.	H	Hm
Stemodia sp.	H	Hm
SOLANACEAE (2/7)
Nicotiana tabacum L.	S	Hm
Solanum asperum Rich.	S	A?
S. baturitense Huber	S	A?
S. crinitum Lam.	S	A?
S. gardneri Sendtn.	S	A?
S. paludosum Moric.	S	A?
S. variabile Mart.	S	A?
STERCULIACEAE (3/4)
Helicteres mollis K. Schum.	S	Hm
H. cf. velutina K. Schum.	S	Hm
Waltheria rotundifolia Schrank	S	Ht
TILIACEAE (1/1)
Luehea sp.	S	Hm
TURNERACEAE (1/2)
Turnera diffusa Willd. ex Schult.	S	Ht
Turnera sp.	H	Ht
VERBENACEAE (2/3)
Lantana camara L.	H	Hm
Lippia gracilis Schau.	H	Hm
L. schomburgkiana Schau.	S	Hm
VIOLACEAE (1/1)
Hybanthus calceolaria (L.) G.K. Schulze.	H	Hm

^* The species are listed by family and in alphabetical order.

^† T, Tree; S, shrub; H, herb; L, liana; E, epiphyte; HC, herbaceous climber.

^‡ Hm, hermaphrodite; Ht, herm. heterostylous; M, monoecious; A, andromonoecious; D, dioecious.

^§ A. V. L. Leite, personal communication.

The frequency of hermaphroditic species was 83 % (122 species), seven of these (or 4·8 % of the total) heterostylous. Monoecy occurred in 9·5 % (14) of the species, followed by andromonoecy in 4·8 % (seven), and dioecy, in only 2·7 %, which was represented by four species: Clusia nemorosa (Clusiaceae) (not really typical of Caatinga areas), Commiphora leptophloeos (Burseraceae), Myracroduon urundeuva and Schinopsis brasiliensis (Anacardiaceae). A high percentage of hermaphroditic species was expected, since this is the most common sexual system in plants (Richards, 1986), and similar percentages have been reported for other tropical ecosystems (Table 2).

Table 2.

Frequency of sexual systems and self-incompatibility (SI) in species of Caatinga as compared with studies of Neotropical and Paleotropical communities

	Hermaphrodite (%)
Studies in tropical communities	Monostylous	Heterostylous	Monoecious (%)	Andromonoecious (%)	Dioecious (%)	SI6 (%)
DRY FORESTS*
Tropical dry forest and scrub vegetation,	78·2	4·8	9·5	4·8	2·71	61·5
‘Caatinga’, Brazil (this study)1					113
					4·34
Savanna, ‘Cerrado’, Brazil (Oliveira and Gibbs, 2000)2	80·0	–	5·0	–	15·0	66
Tropical semi-deciduous forest, Mexico (Bullock, 1985)1	67·2	3·0	14·97	2·0	13·08	76
Tropical deciduous forest, Mexico (Tabla and Bullock, 2002)1	70·2	–	17·6	–	12·3	–
Semi-deciduous forest, Costa Rica (Bawa, 1974; Bawa and Opler, 1975)3	68·0	–	10·0	–	22·0	79·4 (n = 34)9; 54 (n = 130)9
Secondary deciduous forest, Venezuela (Ruiz-Zapata and Arroyo, 1978)1	54·6	9·1	–	13·6	22·7	54·45
RAIN FORESTS*
Atlantic forest, Brazil (Silva et al., 1997)4	78·9	–	8·5	–	12·6	–
Lowland rain forest, Costa Rica (Kress and Beach, 1994)1	70·2	–	12·4	–	17·4	50·9
Lowland rain forest, Costa Rica (Bawa et al., 1985)3	65·5	–	11·4	–	23·1	82·1410
Tropical rain forest, Mexico (Ibarra-Manriquez and Oyama, 1992)3	63·0	–	9·0	–	27·0	–
Moist forest, Barro Colorado, Panama (Croat, 1979)1	76·0	–	15·0	–	9·0	–
Montane cloud forest, Venezuela (Sobrevila and Arroyo, 1982)1	69·453,11	–	–	–	30·553	26·213
	97·4411,12				2·5612	41·7612
Tropical palm swamp, Venezuela (Ramírez and Brito, 1990)1	77·3	–	19·5	–	3·1	30·77
Montane forest, Jamaica (Tanner, 1982)3	68·0	–	11·0	–	21·0	12
Lowland mixed dipterocarp forest, Malaysia (Kato, 1996)5	80·5	–	7·3	–	12·2	–
MARINE INFLUENCED VEGETATION*
‘Restinga’ (coastal vegetation), Brazil (Ormond et al., 1991)1	75·6	–	14·2	–	10·2	–
‘Restinga’ (tropical coastal vegetation), Brazil (Matallana et al., 2005)1	751	–	111	–	141 3513	–

^* Vegetation type as mentioned by the authors.

¹ Including species from all life forms; ² woody species only; ³ trees only; ⁴ trees and shrubs only; ⁵ understorey species only; ⁶ considering hermaphrodite and monoecious species only; ⁷ including 1·6 % of gynomonoecious species and 0·3 % polygamomonoecious species; ⁸ including 0·6 % of gynodioecious species and 0·1 % of androdioecious species; ⁹ data from Bawa (1974);¹⁰ SI frequency value varying: 85·71 %, 82·14 % and 81·48 %;¹¹ including hererostylous and monoecious species according to the authors; ¹² shrubs, herbs and vines only; ¹³ dominant woody species only.

The percentage of monoecious species (9·5 %) agrees with the data of some authors (Table 2), but is higher than the values of 4·5–5 % given for Savanna (‘Cerrado’) communities (Saraiva et al., 1996; Oliveira and Gibbs, 2000), and less than half that found in some tropical evergreen forests. The highest frequency of monoecious species (19·53 %) was reported for a Palm swamp community in Venezuela by Ramírez and Brito (1990) (Table 2), but this value is probably influenced by the predominance of palms in this community. Similarly, in the Caatinga areas studied, the percentage of monoecy can be associated with the high frequency of Euphorbiaceae species, since 12 of the 14 monoecious species belong to this family (Table 1), one of the three most representative families in this ecosystem.

Andromonoecy in the Caatinga (4·8 %) occurs mainly in Solanaceae species (Table 1), a family in which this sexual system is common (Symon, 1979; Coleman and Coleman, 1982; Anderson and Symon, 1989). Data about andromonoecy for other communities are scarce and available only for dry forest communities and with contrasting frequencies (2 % and 13·6 %) (Table 2).

The proportion of dioecious species in the Caatinga (2·7 %) is much lower than in other tropical ecosystems (Table 2), being comparable with the percentage (3 %) recorded by Barbosa (1997) for understorey species of a Brazilian Savanna (Cerrado), and by Ramírez and Brito (1990) for a palm swamp community in Venezuela (3·1 %), which included species of all life forms. However, with respect to Cerrado woody species, Oliveira and Gibbs (2000) recorded 15 % of dioecy. If the percentage of dioecy in the Caatinga for only trees and shrubs is analysed, the value would be 4·3 % or approx. 11 % when only trees are considered. These values are still lower than those reported for most dry and humid forests (when comparisons take into consideration the life forms). The percentages of the sexual systems among diverse ecosystems are summarized in Table 2; these include studies in dry and humid areas, some are for woody species only, others with species of every life form. There are no differences among data between dry and humid communities, and there is a tendency for higher percentages of dioecy in studies restricted to woody species. In general, Renner and Ricklefs (1995) estimate the occurrence of approx. 6 % of dioecious species among angiosperms, a rather higher percentage than in the Caatinga.

Dioecy has been related to small generalist insect pollination (Bawa and Opler, 1975; Bawa, 1980; Bawa et al., 1985) and, in fact, three of the four dioecious species recorded here are pollinated by diverse small insects (Machado and Lopes, 2004). However, Clusia nemorosa has large and very attractive flowers and offers resin as floral reward for female bees of the tribe Euglossini (Lopes and Machado, 1998). In general, the floral traits and pollination systems recorded by Machado and Lopes (2004) in the same Caatinga communities revealed a low percentage of small generalist flowers, which may explain the low occurrence of dioecy.

Dioecy has also been associated with specialized zoochory (Bawa and Opler, 1975; Bawa, 1980; Bawa et al., 1985). This syndrome is present in two of the dioecious species, Commiphora leptophloeus and Clusia nemorosa, which have capsules with seeds covered by a reddish aril, thus probably attracting birds, but the remaining two species, Myracroduon urundeuva and Schinopsis brasiliensis, have dry, abiotic dispersed fruits. Griz and Machado (2001) recorded zoochory for 36 % against 64 % with abiotic dispersed seeds, of the Caatinga species in an area of Pernambuco, a scenario which possibly correlates with the low occurrence of dioecy. A strong association between seed dispersal and dioecy, as well as pollination and other ecological factors was recorded for a rain forest in Mexico (Ibarra-Manriquez and Oyama, 1992).

Another explanation for the reduced percentage of dioecious species in the Caatinga could be the predominance of life forms other than trees (Table 1). Analysis of the growth forms of dioecious angiosperm genera revealed a higher proportion among trees (cf. Bawa and Opler, 1975; Sobrevila and Arroyo, 1982; Bawa et al., 1985; Bullock, 1985; Renner and Ricklefs, 1995), and a comparison of the percentages of dioecious species among canopy, sub-canopy and understorey (Kress and Beach, 1994) recorded the highest values of dioecy among canopy (24·5 %) rather than understorey species (9·8 %). Sobrevila and Arroyo (1982) also recorded a marked difference between the values of dioecy among trees (30·55 %) and shrubs, herbs and vines (2·56 %). In the Caatinga areas, three of the four dioecious species are trees and one (Clusia nemorosa) is a shrub, although this latter species usually occurs as a tree in the Brazilian Amazonian and Atlantic forests. Thus, as mentioned above, the percentage of dioecy among trees in the present study is 11 % (three out of 28 tree species), or 14·3 % if Clusia nemorosa is included.

Breeding system studies

The 39 species for which breeding system studies are available (Table 3) belong to 13 families, including the three most common in Caatinga communities, Cactaceae, Euphorbiaceae and Leguminosae. Analysis revealed a high percentage of self-incompatible (SI) species (61·5 %) represented by 24 species, including the six heterostylous ones [Auxemma glazioviana, A. oncocalyx, Cordia globosa, C. leucocephala (Boraginaceae), Melochia tomentosa and Waltheria rotundifolia (Sterculiaceae)]. The percentage of self-compatibility (SC) was 38·5 %, recorded for 15 species of which seven species set fruit after spontaneous self-pollination. Agamospermy was not recorded in the Caatinga species studied. A similar low (or null) level of agamospermy was recorded for other tropical communities, both dry (Bawa, 1974; Ruiz-Zapata and Arroyo, 1978; Bullock, 1985) or humid (Ramírez and Seres, 1994). The high frequency (15·38 %) of agamospermy in a morichal community was associated with inappropriate pollinators (Ramírez and Brito, 1990). These authors also drew attention to the incidence of agamospermy in some Melastomataceae (see review by Renner, 1989), a family which was not represented in the present study.

Table 3.

Data compiled for breeding system of 39 species occurring in Caatinga areas in north-eastern Brazil

Family and species	Habit	Cross-pollination	Spontaneous self	Manual self	Agamospermy	Control	Breeding systems*	Source
AGANTHACEAE
Harpochilus neesianus Mart.	Shrub	–	–	–	–	12·17 %	SI?	Vogel et al., 2004
Ruellia asperula (Nees) Lindau	Herb	(16/12) 75·0	(27/10) 37·0	(17/13) 76·5	(10/0) 0	(17/6) 35·5	SC	Machado, 1990; Machado and Sazima, 1995
R. aff. paniculata L.	Herb	(10/10) 100·0	(20/20) 100·0	(10/9) 90·0	(10/0) 0	Ca. 90·0 %	SC	Machado, 1990
BORAGINACEAE
Auxemma glazioviana Taub.	Tree	Intermorph (261/116) 45·0†	(3393/0) 0	(321/0) 0	(176/0) 0	(4454/354) 8·0	SI	Silva and Machado, 1997
A. oncocalyx (Fr. All.) Taub.	Tree	Intermorph (560/150) 27·5†	(5479/0) 0	(287/0) 0	(118/0) 0	(14140/217) 2·0	SI	Silva and Machado, 1997
Cordia globosa (Jacq.) Kunth	Shrub	Intermorph (20/15) 75·0†	(20/0) 0	0	0	Ca.70·0 %	SI	Machado, 1990
C. leucocephala Moric.	Shrub	Intermorph (20/1) 7·5†	(20/0) 0	0	0	Ca.5·0 %	SI	Machado, 1990
CACTACEAE
Melocactus zehntneri (Britton & Rose) Luetzelburg	Herb	(15/12) 80·0	(20/17) 85·0	(10/9) 90·0	(10/0) 0	(20/19) 95·0	SC	Locatelli and Machado, 1999
Opuntia palmadora Britton & Rose	Shrub	(30/19) 63·0	(30/21) 70·0	30/23) 76·0	(20/0) 0	(30/25) 83·0	SC	Locatelli and Machado, 1999
CAPARACEAE
Capparis hastata L.	Shrub	(23/19) 82·6	(42/0) 0	(42/14) 33·3	(32/0) 0	(141/56) 39·7	SC	Primo, 2004
C. jacobinae Moric ex Eichler	Shrub	(72/51) 17·2	(33/0) 0	(105/1) 0·9	(32/0)	(99/21) 21·2	SI	Primo, 2004
COMBRETACEAE
Combretum leprosum Mart.	Liana	(50/8) 16·0	(50/0) 0	(50/0) 0	(50/0) 0	(100/19) 19·0	SI	Quirino and Machado, 2001
C. pisonioides Taub.	Shrub	(50/10) 20·0	(50/0) 0	(50/0) 0	(30/0) 0	(40/4) 10·0	SI	Quirino and Machado, 2001
CONVOLVULACEAE
Ipomoea asarifolia asarifolia (Ders.) Roem. & Schult.	Liana	(30/18) 60·0	(30/0) 0	(30/1) 3·0	(30/0) 0	(30/23) 76·0	SI	Piedade, 1998; Kiill and Ranga, 2003
I. incarnata Choisy	Liana	(30/22) 73·0	(30/0) 0	(30/2) 6·0	(30/0) 0	(30/25) 83·0	SI	Piedade, 1998
I. bahiensis Willd. ex Roem. & Schult.	Liana	(30/17) 56·0	(30/0) 0	(30/0) 0	(32/0) 0	(31/21) 67·0	SI	Piedade, 1998
I. brasiliana Meisn.	Liana	(30/7) 23·0	(30/1) 3·0	(30/0) 0	(32/0) 0	(35/10) 28·0	SI	Piedade, 1998
I. martii Meisn.	Liana	(30/0) 0	(30/0) 0	(30/0) 0	(30/0) 0	(30/0) 0	SI?	Piedade, 1998
Jacquemontia multiflora (Choisy) Hallier f.	Liana	(30/18) 60·0	(30/1) 3·0	(30/10) 33·0	(30/0) 0	(31/19) 60·0	SC	Kiill and Ranga, 2000b
Merremia aegyptia L. Urban.	Liana	(30/19) 63·0	(30/10) 33·0	(30/13) 43·0	(30/0) 0	(30/25) 83·0	SC	Kiill and Ranga, 2000a
EUPHORBIACEAE
Jatropha gossypifolia L.	Shrub	(40/33) 82·5	–	(40/30) 75·0‡	–	(100/95) 95·0	SC	Ormond et al., 1984
J. mollissima (Pohl.) Baill.	Shrub	(20/19) 95·0	–	(20/13) 65·0‡	(20/0) 0	(20/17) 85·0	SC	Santos et al., 2005
J. mutabilis (Pohl.) Baill.	Shrub	(20/16) 80·0	–	(20/12) 60·0‡	(20/0) 0	(20/15) 75·0	SC	Santos et al., 2005
LEGUMINOSAE
Caesalpinia calycina Benth.	Shrub	(19/5) 26·3	0	(23/0) 0	–	(3083/393) 12·75	SI	Lewis and Gibbs, 1999
C. pluviosa DC. var. sanfranciscana G.P.Lewis	Tree	–	–	–	–	(100/4) 4·0	SI?	Lewis and Gibbs, 1999
Chamaecrista cytisoides (Collad.) Irwin & Barnely	Shrub	(30/12) 40·0	(37/0) 0	(30/8) 26·6	–	(106/56) 52·8	SC	Silva, 2004
Ch. ramosa (Vogel) var. mollissima (Benth.) Irwin & Barneby	Shrub	(30/18) 60·0	(57/0) 0	(30/15) 50·0	–	(110/22) 20·0	SC	Silva, 2004
Gliricidia sepium (Jacq.) Steud.	Tree	(31/16) 51·6	(36/0) 0	(30/0) 0	(30/0) 0	(30/9) 30·0	SI	Kiill and Drummond, 2001
MALPIGHIACEAE
Byrsonima gardneriana Juss.	Shrub	(30/28) 93·3	(30/0) 0	(30/0) 0	(30/0) 0	(129/68) 52·7	SI	Bezerra, 2004
Stigmaphyllom paralias Juss.	Shrub	(30/26) 88·66	(30/0) 0	(30/0) 0	(30/0) 0	(45/38) 84·4	SI	Bezerra, 2004
MALVACEAE
Pavonia martii Mart. ex Cola	Herb	(10/10) 100·0	(20/20) 100	(10/10) 100·0	(10/0) 0	98·0 %	SC	Machado, 1990
RHAMNACEAE
Ziziphus joazeiro Mart.	Tree	(29/9) 31·0	(30/0) 0	(40/7) 15·5	–	(96/15) 15·6	SC	Pinheiro et al., 1991
SCROPHULARIACEAE
Angelonia bisaccata Benth.	Shrub	(20/18) 90·0	(10/0) 0	(10/0) 0	(10/0) 0	Ca. 65·0 %	SI	Machado, 1990; Vogel and Machado, 1991
A. cornigera Hook.	Herb	(20/17) 85·0	(15/0) 0	(21/0) 0	(20/0) 0	Ca. 60·0 %	SI	Machado et al., 2002; this study
A. hirta Cham.	Herb	(20/20) 100·0	(20/0) 0	(20/0) 0	(10/0) 0	Ca. 65·0 %	SI	Machado, 1990; Vogel and Machado, 1991
A. hookeriana Gardn.	Shrub	(15/14) 93·3	(10/0) 0	(10/0) 0	(10/0) 0	Ca. 65·0 %	SI	Machado, 1990; Vogel and Machado, 1991
A. pubescens Benth.	Herb	(20/20) 100·0	(20/20) 100·0	(20/20) 100·0	(20/0) 0	Ca.100·0 %	SC	Machado, 1990; Vogel and Machado, 1991
STERCULIACEAE
Melochia tomentosa L.	Shrub	Intermorph (110/108) 98·2†	(120/0) 0	0	0	Ca. 90·0 %	SI	Machado, 1990
Waltheria rotundifolia Schrank	Shrub	Intermorph (20/12) 60·0†	(20/0) 0	(10/0) 0	(10/0) 0	Ca. 65·0 %	SI	Machado, 1990; this study

The values are (number of flowers/fruits) %.

^* SC, Self-compatible; SI, self-incompatible.

^† Intramorph crosses did not result in fruits.

^‡ Result refers to geitonogamy since the species is monoecious.

The high frequency of SI species in the Caatinga is similar to records by Oliveira and Gibbs (2000) for a Cerrado community in Brazil, and also to other dry (Bawa, 1974; Bullock, 1985) and humid (Bawa et al., 1985) tropical forest communities (Table 2). Although SI has been reported as predominant in tropical communities (Bawa et al., 1985, Jaimes and Ramírez, 1998), higher frequencies of SC in relation to SI species were reported by Sobrevila and Arroyo (1982) for a Venezuelan montane forest (26·21 % of SI considering trees only or 41·76 % of SI for shrubs, herbs and vines), by Ramírez and Brito (1990) for a palm swamp community also in Venezuela (30·77 % of SI species) and by Tanner (1982) for a Jamaican montane forest (12 %) (Table 2).

Although the frequency of SI species was similar to other forests, the level of SI plus the frequency of dioecy in the Caatinga gives a smaller frequency of obligate outcrossing species for this community in comparison, for example, with the Cerrado studied by Oliveira and Gibbs (2000) with 66 % SI species and 15 % with dioecy, a total of 81 % of obligatory outcrossing species. However the frequency of total obligate outcrossing species in the Caatinga (61·5 SI + 2·7 dioecy), even with a low level of dioecy, is still higher than some humid ecosystems showed in Table 2, such as a tropical palm swamp (Ramirez and Brito, 1990) and a montane cloud forest (Sobrevila and Arroyo, 1982), both in Venezuela, and a Jamaican montane forest (Tanner, 1982).

With regard to the habit of the 39 species with breeding system data (Table 3), high frequencies of SI species predominate in all woody life forms (80 % of the tree species, 63·2 % of the shrubs, 75 % of the lianas), but less so in herbs (28·6 %). The herbaceous annual habit has been associated with self-compatibility (Stebbins, 1950), and it is of interest to note that the two herbaceous species with SI in the present survey are in fact perennial species.

The relatively small sample of species in this study indicates that high levels of SI seem to predominate in Caatinga species independent of taxonomic affinities, thus making cross-pollination services of great importance for this ecosystem, as proposed for the Cerrados (Oliveira and Gibbs, 2000) and tropical rain forests (Bawa et al., 1985). More studies concerning the breeding and incompatibility systems of Caatinga species are needed to verify what types of reproductive mechanisms occur in plant species of this ecosystem.

We are grateful to P. E. Gibbs (University of St Andrews, Scotland) and S. Bullock (University of San Diego, USA) for critically reading the first drafts of the manuscript and also for kindly improving the English, also to S. Vogel (University of Vienna, Austria) for fruitful discussions. The Curators and Taxonomists of the Herbaria UFP, IPA and PEUFR and the taxonomists F. Agra (UFPB), G. S. Baracho (UFPE), L. P. Felix (UFPB), I. B. Loiola (UFRN), L. Paganucci (UEFS), R. Pereira (IPA), J. Semir (UNICAMP), J. A. Siqueira-Filho (UFPE), V. Souza (Esalq-USP) and D. Zappi (Kew Gardens) for identifying most of the plant species. Three referees (P. Gibbs, P. Oliveira and one anonymous) for several relevant improvements in the manuscript. Also to A. V. L. Leite for unpublished data, and the owners/administrators of the study sites for permission to carry out our studies at areas under their care. The UFPE, CNPq, FACEPE, CAPES, for essential financial support.

Author notes

1Departamento de Botânica—CCB—Universidade Federal de Pernambuco, 50372-970, Recife-PE, Brazil and 2Departamento de Botânica—IB—Universidade Estadual de Campinas, 13083-970, C.P. 6109, Campinas-SP, Brazil