Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Reproductive strategies, relichenization and thallus development observed in situ in leaf-dwelling lichen communities
William B. Sanders,
Departamento de Botânica, Centro de Ciências Biológicas, Universidade Federal de Pernambuco, Recife–PE, and the University Herbarium, University of California, Berkeley, CA 94720-2465 USA; Present address: Centro de Ciencias Medioambientales, CSIC, Calle Serrano 115 bis, E–28006 Madrid, Spain;
Author for correspondence: William B. Sanders Tel: +34 91 7452500 ext. 274 Fax: +34 91 5640800 Email: firstname.lastname@example.org
Here we use the term ‘algae’ to refer to the broad, polyphyletic assemblage of photosynthetic life forms exclusive of the embryophytes, as do contemporary phycology textbooks. Almost all algae found in lichen associations are either green (chlorophytes) or blue-green (cyanophytes/cyanobacteria); two lichens with heterokontophyte algal symbionts are also known ( Tschermak-Woess, 1988 ). The widely used term ‘phycobiont’ is applied to all these algal symbionts.
Author for correspondence: William B. Sanders Tel: +34 91 7452500 ext. 274 Fax: +34 91 5640800 Email: email@example.com
• Suppositions about lichen reproductive strategies were investigated and elusive early stages of lichen ontogeny documented in a foliicolous lichen community.
• Plastic coverslips attached to supportive netting were placed among foliicolous lichen communities within a neotropical lowland forest. The germination and development of diverse lichen propagules colonizing the coverslips were studied with light microscopy.
• Foliicolous lichens were observed to begin development from lichenized vegetative propagules, aposymbiotic fungal spores, fungal spores dispersed together with attached phycobionts, and diahyphae. Aposymbiotically dispersed spores and diahyphae were capable of associating with compatible phycobionts encountered upon the substratum, following germination.
• Many developing thalli produced characteristic structures (discoid isidia, thalline setae, pycnidia, etc.) which permitted their recognition as typical members of the foliicolous lichen community. Thalline setae in Tricharia were produced upon the prothallus, and subsequently incorporated into the thallus proper by advance of the lichenized thallus margin. Tricharia and other members of the Gomphillaceae showed a distinctive organization of symbionts in thallus growth, whereby the unicellular green phycobiont cells were positioned at the tips of advancing fascicles of mycobiont hyphae. In Coenogonium sp., branching filaments of the phycobiont Trentepohlia grew along prothallic paths initiated by the mycobiont.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
About one-fifth of the known species of fungi form lichen symbioses with algae (Kirk et al., 2001).* The lichen fungi enclose algal cells within an often plant-like body (Sanders, 2001b) in which the alga supplies the fungal tissues with carbon fixed by photosynthesis (Smith, 1980). For these fungi the symbiosis is, as far as is known, obligatory for the completion of their life cycles. The lichen fungus does not form stable symbioses with a broad range of algae; rather, each fungal species appears to show a fairly high degree of selectivity in its choice of partners (Ahmadjian & Jacobs, 1981; Beck et al., 1998; Paulsrud et al., 1998, 2000; Kroken & Taylor, 2000). Many lichens appear to maintain the partnership in reproduction by means of vegetative propagules, such as soredia, isidia or other types of thallus fragments, which disperse both symbionts together (Bailey, 1976; Schuster et al., 1985; Scheidegger, 1995). On the other hand, a great number of lichens lack recognizable vegetative propagules. They produce fungal ascospores (or basidiospores), presumably resulting from sexual processes, and/or asexual spores such as conidia. In some lichens, phycobiont cells may be present within or upon the fungal spore-producing tissues, where they can be dispersed together with the liberated spores (Bertsch & Butin, 1967). However, in most cases, the fungal spores seem to be dispersed aposymbiotically; upon germination, the fungus would need to re-encounter and lichenize a suitable algal symbiont from the environment. Encounter of a suitable alga is often thought to be problematic because of the reportedly infrequent occurrence of some important phycobionts, such as Trebouxia species, in the free-living state. However, the question of phycobiont availability in nature requires further study. Although a few notable works have shown some apparent stages of lichen resynthesis (Ward, 1884; Werner, 1931; Jahns et al., 1979; Garty & Delarea, 1988), very little is known about the occurrence, procedure and relative success of lichen reproductive strategies. Some lichenologists have even doubted whether relichenization does occur in nature (Lamb, 1959). Others have chosen alternative approaches, ‘… since the process of lichenization itself cannot be observed in nature …’ (Beck et al., 1998). While increasing attention is being focused upon acquisition and loss of the lichen mode of nutrition in fungal evolution (Gargas et al., 1995; Lutzoni et al., 2001), we remain largely ignorant of how and to what extent such events might occur regularly in the life cycles of these organisms. The subsequent processes involved in lichen thallus organization and development are also poorly understood. What is needed is an adequate system for study of the early, microscopic stages of lichen ontogeny in situ. The crustose lichen communities that occur on the surfaces of leaves in humid tropical forests (Santesson, 1952; Lücking, 2001) are ideal subjects for such studies. These foliicolous (i.e. leaf-dwelling) lichens will develop in situ on artificial substrata, such as plastic (Sipman, 1994; Lücking, 1998), and may colonize transparent materials that permit their direct observation with light microscopy (Sanders, 2001a). Furthermore, because they are adapted to a short-lived natural substratum, foliicolous lichens develop rapidly.
In a preliminary study of foliicolous lichen colonization in situ, some evidence of lichenization as well as symbiont codispersal was observed on glass slides taped to a plastic support within a lowland tropical forest (Sanders, 2001a). However, the low diversity of lichens observed on the glass slides and their frequent failure to develop beyond early stages suggested that the substratum and/or microenvironment were suboptimal for development of foliicolous lichens. This situation also raised some questions about whether the observed associations indeed represented development of typical foliicolous lichen thalli. In the present study, these problems were solved by using a clear plastic substratum attached without tape or glue directly over the surfaces of leaves. A variety of typical foliicolous lichens developed readily on the experimental substratum. Their varied reproductive strategies and some of the patterns of subsequent thallus development are documented here.
Materials and Methods
The present study was carried out from March 2000 to March 2001 at the Parque Estadual de Dois Irmãos, Recife, Pernambuco, Brazil. The park conserves a remnant of the native Atlantic forest vegetation. A floristic treatment of the foliicolous lichens of coastal Pernambuco (Cáceres, 1998; Cáceres et al., 2000) cites numerous collections from this field site. Plastic screening with a mesh of about 2 mm square was cut into strips 3–4 cm wide and 20–35 cm long. Plastic microscope cover slips (20 × 20 mm) were fastened to the strips of screening by inserting their corners into small diagonal slits cut into the mesh. A total of 31 strips with five to nine cover slips per strip were prepared. The strips were tied, with the cover slips facing upward, to the upper surfaces of leaves of Bactris sp., a common understory palm which typically bears well-developed foliicolous lichen communities (Fig. 1a). Bactris plants were selected from several different localities throughout the reserve. Plants with leaves already colonized by lichens were chosen, as this was an indication of microenvironmental conditions favorable to development of these organisms. At intervals of 3–6 wk, some of the cover slips were removed for microscopic examination. Debris and microorganisms were wiped off the lower surface of the cover slips, which were then inverted and mounted with distilled water onto clean glass microscope slides. In a few instances, a clearer image was obtained by placing the colonized cover slip face up on the microscope slide, with a glass cover slip wet-mounted over it. Attempts to make sequential observations on the same individuals by returning cover slips to the field were not successful; exposure to the intense light of the microscope lamp was believed to be deleterious to the algal symbionts. After about 6 or 7 months from the time of placement in the field, most of the cover slips became heavily covered with foliicolous lichens and other organisms, as well as organic debris, impeding clear microscopic observation of the behavior of individual organisms present. Most of the observations presented here are therefore from cover slips withdrawn for observation during the first 6 months after their placement in the field.
Within 3–5 wk after placement, many cover slips were colonized by germinating fungal spores, as well as several types of chlorococcalean green algal cells. Among the latter were cells of Trebouxia, identified by their characteristic chloroplast (Fig. 1b). They were most often scattered singly upon the substratum, occasionally in groups of two or three, free of association with fungi. These aposymbiotic Trebouxia cells were occasionally observed in division (Fig. 1c). The discoid filamentous green alga Phycopeltis, an important symbiont in foliicolous lichens, occurred abundantly in the free-living state. The first fully recognizable lichens were minute thalli of Phyllophiale, which developed from characteristic disc-shaped isidia (Sanders, 2001a). Other vegetative lichen propagules, containing both fungal and algal symbionts, were commonly observed germinating on the substratum. The peripheral fungal cells produced hyphae that grew outward, often associating with free algal cells encountered on the substratum (Fig. 1d). Thus, the possession of a phycobiont in proximal portions of the young lichen did not appear to preclude incorporation of additional algae from the external environment by the advancing hyphae. Some dispersed fragments of lichen thalli of the family Gomphillaceae, consisting mainly of single hyaline setae, were seen propagating from remnants of the alga-containing thallus at the base of the seta (Fig. 1e). Soredia and similar lichenized propagules were also observed germinating; these probably represent lichen taxa which normally colonize branches and trunks.
Large, muriform (both longitudinally and transversely septate) fungal spores, most probably representing species of Gomphillaceae or Ectolechiaceae, produced a great number of germination hyphae from their unit cells, giving rise to a dense, radiating mat of mycelium. In several instances the hyphae could be seen surrounding trebouxioid algal cells encountered on the substratum, incorporating them into distinctive structures (Fig. 1f). Many of the enclosed algal cells were dividing. Fusiform, transversely septate fungal spores (Trichotheliaceae) were frequently observed with germination hyphae emerging from the terminal cells at opposite ends of the spore. The hyphae grew towards algal cells, especially thalli of Phycopeltis. The hyphae grew between and around the filamentous branches of Phycopeltis, surrounding the alga and producing hyphal branches, which radiated out over the substratum (Fig. 1g). The alga appeared to remain healthy.
Germinating lichen spores, possibly belonging to Porina rubentior or a related taxon (Trichotheliaceae), were also seen growing toward and apparently contacting already lichenized Phycopeltis within young Phyllophiale thalli (Fig. 1h). However, it was not possible to confirm whether any of the lichenizing hyphae associated with the alga originated from these germinated spores.
Some colonizing lichens began their development from fungal spores dispersed together with their algal symbiont. Filiform, septate spores wrapped around green algal cells were repeatedly observed in germination (Fig. 1i). These were campylidial conidia of Tapellariopsis, which typically has phycobiont cells present in the chambers where conidia are produced. The dispersed spores and algal cells were frequently observed in large clusters which merged in growth to produce a single continuous thallus. Large muriform ascospores (possibly Gyalectidium filicinum or Gyalideopsis vulgaris) with algal cells attached to their surfaces were also seen germinating on the coverslips (Fig. 1j). Their germ hyphae enveloped the attached algal cells and radiated out over the substratum.
The specialized propagular diahyphae characteristic of the Gomphillaceae and apparently unique to this group of fungi were also frequently seen. The diahyphae consisted of wefts of hyphae or chains of short fungal cells, which were generally dispersed as a unit. Numerous individual cells of the diahyphal chains produced germination hyphae (Figs 1k and 2a). Mycelia developing from germinating diahyphae contacted and incorporated trebouxioid algae encountered on the substratum (Fig. 2b).
Germination hyphae and prothallic hyphae of young thalli also frequently contacted types of free-living algae clearly different from their phycobiont. In some cases, the contacting hyphae simply grew over or around the foreign algae, while in other cases the alga was encircled, its cells invaded and apparently destroyed by the hyphae (Fig. 2c).
While the fate of individual interactions observed could not be followed directly over time, a continuum of later stages of lichen thallus formation occurred on the experimental substratum. Formation of ascocarps was not observed within the study period, although some thalli did produce pycnidia (see Fig. 3g). Many of the larger thalli developed characteristic structures, such as setae (Fig. 2d) or discoid isidia, which confirmed that these lichens were typical members of the foliicolous community. Dark-pigmented, tapered setae, characteristic of the lichen genus Tricharia, were initiated upon the prothallus of hyphae extending beyond the alga-containing margin of the thallus. The earliest stage of seta initiation was evident as a distinct, circumscribed pad of closely appressed fungal cells which were initially hyaline (Fig. 2e). These prothallic cells produced a dark pigment (Fig. 2f) prior to vertical growth of the seta (Fig. 2g). The base of the seta was eventually surrounded by thalline tissue as the lichenized, alga-containing margin advanced beyond the base of the seta (Fig. 2h). Setae in various stages of development are visible on the Tricharia thallus shown in Fig. 2d.
A distinctive type of symbiont organization was evident at the margin of young Tricharia thalli, as well as in related taxa in the Gomphillaceae that produced hyaline setae. The phycobiont cells were positioned at the tips of a radially advancing fascicle of mycobiont hyphae (Fig. 2i). In many lobes these algal cells could be observed in active division (Fig. 3a). Proximally from the margin, hyphae diverging upward directed phycobiont cells into an algal layer forming above the hyphal fascicles. The positioning of the algal cells at hyphal tips appeared to be organized at an early stage of development (Fig. 3b).
A different type of organization was observed in a crustose Coenogonium (= Dimerella; Lücking & Kalb, 2000). This lichen produced an asterisk-shaped thallus containing the filamentous Trentepohlia as phycobiont. The earliest stages observed were dense clusters of the symbionts which appeared to result from some sort of vegetative propagule or fragment (Fig. 3c). The fungal prothallus of this lichen did not grow uniformly over the substratum; rather, the radial prothallic hyphae were grouped together, extending in several spokes or rays. Filaments of Trentepohlia advanced along these prothallic rays (Fig. 3d), giving rise to a thallus with deeply incised lobes. The fungal hyphae made somewhat intermittent lateral contacts with the algal filaments, often passing between separate algal filament branches (Fig. 3e). Some larger thalli produced pycnidia, which bore characteristic one-septate bacillar conidia.
Young lichen areoles growing in proximity often appeared to merge during development to form a continuous thallus (Fig. 3f). In other instances, however, a conspicuous zone of interaction developed where the prothalli of adjacent young thalli came into contact and sometimes fused (Fig. 3g). Within this zone, the hyphae sometimes showed extensive rebranching, septation, swelling, and some apparent vacuolation (Fig. 3h). In some cases, a brownish discoloration of the hyphae within the contact zone was apparent. Between neighboring thalli of obviously different lichen species, inhibition was often apparent in the locally restricted growth of the phycobionts of the adjacent thalli (Fig. 3i).
The results indicate that dispersal of (1) lichenized vegetative propagules and fragments, (2) aposymbiotic fungal spores, (3) fungal spores together with attached phycobionts and (4) diahyphae, all represent fully functional strategies of early colonization in foliicolous lichens. Early colonization was characterized by relatively large fungal spores or propagules, usually well-provided with food reserves in the form of large oil droplets. The fungus generally germinated from numerous peripheral points and spread in all directions, resulting in rapid establishment. Diahyphae germinated as a unit rather than dissociating into component cells. The dispersion of larger, well-provisioned propagules capable of mobilizing greater resources may permit a more rapid colonization of the ephemeral leaf substratum. The thin-walled ascospores which predominate among foliicolous lichens have also been seen as an adaptation for rapid development (Santesson, 1952). Large ascospores, however, appear to be a general feature of tropical rainforest lichens regardless of substratum; the significance of this feature remains to be fully clarified (Sipman & Harris, 1989).
Both aposymbiotic spores and diahyphae were capable of reassociating with algal symbionts encountered upon the substratum following germination. The associations depicted in Figs 1f,g and 2a,b appear to represent the earliest stages in reestablishment of the lichen symbiosis. The fungi in Fig. 1g,h are members of the Trichotheliaceae, the most abundant and diverse group among foliicolous lichens; these fungi appear to rely chiefly on aposymbiotically dispersed spores for reproduction (Santesson, 1952; Aptroot & Sipman, 1993; Lücking, 1996). The Gomphillaceae (Figs 1e,f,k and 2a,b, and possibly Fig. 1f,j) are the most diverse group of foliicolous lichens and the second most abundant next to the Trichotheliaceae. They appear to be highly adapted for both symbiotic and aposymbiotic modes of dispersal (Vězda, 1979; Vězda & Poelt, 1987; Lücking, 1997; Ferraro et al., 2001).
The specific identities of the algae have not been determined. Accurate identification of lichen phycobionts requires isolation and culture in the laboratory or molecular tools, and very few phycobionts of foliicolous lichens have been investigated (Tschermak-Woess, 1988). Preliminary molecular analyses indicate that Trebouxia and/or close relatives are the principal phycobionts in foliicolous members of the Gomphillaceae and Ectolechiaceae (R. Lücking, unpublished); the Trichotheliaceae have trentepohliaceous phycobionts, in particular the alga Phycopeltis. Representatives of these algal genera were observed on the cover slips both in the free-living state and in various stages of lichenization with germinating fungal propagules. The incorporation of additional external algae by hyphae of already lichenized propagules and thalli suggests that algal populations within lichen thalli might not be genetically uniform, or that their genetic composition might change in the course of development. The latter situation has been reported in the lichen Diploschistes muscorum (Friedl, 1987). Other reports have suggested that the converse process, substitution of the mycobiont, may also occur; lichen fungi might associate with already lichenized algae, replacing the original lichenizing fungus of vegetative propagules (Ott, 1987) or mature thalli (Poelt, 1974; Zehetleitner, 1978; Hawksworth et al., 1980). Such ‘theft’ of phycobionts might occur where microenvironmental conditions are suboptimal for development of a given lichen but more favorable to an invading lichen fungus. Because of the difficulty in determining from which fungus the lichenizing hyphae originate, the present study could not provide clear evidence for phycobiont theft. However, the observation that germinating spores will grow toward and apparently contact lichenized Phycopeltis thalli (Fig. 1h) at least suggests that the initial events postulated for such a process do indeed occur in this lichen community.
The results indicate that for many lichen fungi the symbiosis is interrupted and reacquired in the course of their life cycle. The existence of an aposymbiotic phase provides the opportunity for new nutritional experiments upon which natural selection may act. Selection may favor the germinating lichen fungal propagule which can take advantage of an alternative substrate when compatible algae are not available, leading to the eventual loss of the lichen mode of nutrition in some lineages. Conversely, in the case of nonlichen fungi, selection pressures may favor those germinating spores capable of making use of aposymbiotic lichen algae encountered in the absence of the fungus's usual substrate. New acquisition of the lichen habit by a fungus might also occur by association with algae not previously involved in lichen symbioses. However, it is interesting to note that separate lineages of fungi that acquired lichen symbiosis independently in the Basidiomycota (Omphalina, Dictyonema and Multiclavula) have phycobionts (Coccomyxa and Scytonema) of the same genera as those associated with Ascomycota (Tschermak-Woess, 1988). This suggests that new acquisition of lichen symbiosis by fungi might be more likely to occur with algae already adapted (or predisposed) to lichen symbiosis.
Developing lichen thalli revealed strikingly different patterns of organization by which mycobiont and phycobiont growth were coordinated. In several members of the Gomphillaceae (Tricharia and related taxa), the chlorococcoid phycobiont cells were organized at the periphery of the young lobes, their spherical cells positioned at the tips of parallel, longitudinally appressed hyphae growing outward from the thallus. This type of organization appears to provide a very simple mechanism by which growth of the mycobiont directly distributes the dividing phycobiont cells to the youngest parts of the radially expanding thallus. Inasmuch as the algal cells and their division products remain positioned at the tips of the advancing hyphae and their branches, apical growth of the hyphae would be sufficient to move the algal cells. Thus, it is not necessary to postulate nonapical or diffuse growth of fungal cells to explain algal distribution in such a system, although such nonapical growth might indeed occur. Various investigators have considered the question of how unicellular phycobionts become distributed throughout the lichen thallus in the course of mycobiont growth. Nienburg (1917) described localized, ephemeral bundles of Schiebehyphen (‘pushing hyphae’) in the lichen Pertusaria which appeared to direct isolated algal cells from within already differentiated, stratified portions of the thallus towards the prothallus at the periphery. In an examination of symbiont distribution mechanisms in Parmelia saxatilis, Greenhalgh & Anglesea (1979) did not observe any structures resembling Nienburg's Schiebehyphen. They attributed algal distribution to the repeated penetration of Trebouxia spore packets and subsequent separation of the daughter cells. A comparable process of separation and short-distance shifting of algal division products was proposed by Honegger (1987) to account for phycobiont distribution in complex lecanoralean lichen thalli. These mechanisms may represent more sophisticated, less obvious variants of the organizational system observed here in the simple crustose thalli of gomphillaceous lichens. By contrast, in Coenogonium, where both the fungus and its algal symbiont are filamentous, no comparable means of mechanical distribution of the phycobiont was evident. The algal filaments were seemingly autonomous in their growth and branching, yet their direction of development rather strictly followed the rays of prothallic hyphae laid down by the mycobiont. Coordination of symbiont growth in this lichen might depend more closely on substances released by the mycobiont.
Observation of seta development in Tricharia indicates that, while these structures often occur centrally on the lichenized, alga-containing thallus, they are actually formed peripherally upon the fungal prothallus. Structural organization of the thallus in this lichen thus appears to be overwhelmingly driven by the mycobiont. At the mycelial periphery the fungus initiates highly differentiated setae which tower above the thallus; fascicles of radially advancing hyphae move the phycobiont cells centrifugally to cover the prothallus with an organized layer of algal cells, surrounding the bases of the setae. Considering their size relative to the diminutive thallus, the setae would appear to represent a substantial investment of the lichen's resources, yet their function is unclear. Homologous structures often borne alongside them, the hyphophores, serve in production of the diahyphal propagules. Usually, however, the setae greatly outnumber the hyphophores, and are frequently numerous on thalli lacking hyphophores altogether. Some evidence of setae functioning as detachable propagules was observed in this study (Fig. 1e). However, those were shorter, hyaline setae of another taxon in the Gomphillaceae, perhaps Echinoplaca. The long black setae of Tricharia were not observed to disperse upon the experimental substratum and their germination capacity is unknown.
Interaction between mycobionts of neighboring thalli were also observed. Although such interactions have previously eluded microscopic study, there is reason to expect that neighboring lichen fungi interact in ways similar to nonlichen fungi. Researchers studying nonlichen fungi have characterized antagonistic/inhibitory interactions between nearby hyphae of competing species. These competitive interactions may include inhibition by depletion of resources, antagonism by antibiosis, or a cytological interference response produced by intimate contact between hyphae of competing species (Ikediugwu, 1976a,b; Rayner & Webber, 1984). Between the two neighboring thalli shown in Fig. 3i, whose different phycobionts clearly indicate that the lichens are of different species, a concave zone of mycelial avoidance is present in the otherwise circular perimeter of the Trentepohlia-containing lichen, where it approaches a neighboring lichen containing a chlorococcoid phycobiont. This inhibition is most likely produced by antibiosis, to which both the fungal prothallus and the Trentepohlia phycobiont appear to respond. In contrast to mycelial antagonism between competing species, mycelia of the same or closely related species may express somatic incompatibility, by which genotypically different individuals reject nonself cytoplasm following hyphal fusion (Aylmore & Todd, 1984; Rayner et al., 1984). The dark lines and inhibition zones frequently visible between adjacent crustose lichen thalli (including foliicolous taxa) have been attributed to somatic rejection responses (Rayner et al., 1984; Rayner, 1991; Clayden, 1997), although duplicating these interactions with axenically cultured mycobionts has been problematic (Dyer et al., 2001). The unusual morphology of the prothallic hyphae within the contact zone between certain thalli (Fig. 3g,h) may represent such a response. Hyphae from the adjacent lichens appear to be fused in the contact zone; they differ strikingly from prothallic hyphae elsewhere in their high degree of septation, swelling, and vacuolization, as well as in the extensive rebranching from hyphae peripheral to the contact zone.
Although leaf-dwelling lichens are evidently adapted for colonization of a specific type of substrate and microhabitat (Lücking, 2001), there is little reason to believe that they differ fundamentally from lichens of other substrates in their biology and development. The same basic types of propagules occur in other lichens of different habitats and substrates; the development of these propagules presumably follows similar principles. The ease with which foliicolous lichen development can be studied in situ should prove to be of great advantage in future experimental investigations of lichen biology and ecology.
The first author would like to thank the Federal University of Pernambnco (UFPE) for the opportunity to fence as visiting professor at that Institution, and Dr Isabelle Tavares for her counsel and generosity. The kind cooperation of Luis Carlos Mafra and Alexandre Albuquerque of the Parque Estadual de Dois Irmãos, Recife, is gratefully acknowledged. Figures were composed using facilities at the Scientific Visualization Center, UC Berkeley, with technical advice kindly provided by Dr W. P. Chan. The manuscript benefited from critical review by Drs Isabelle Tavares, Richard L. Moe, and two anonymous referees.