Effects of Road Clearings on Movement Patterns of Understory Rainforest Birds in Central Amazonia SUSAN G. W. LAURANCE,*t? PHILIP C. STOUFFER,:}:! AND WILLIAM E LAURANCEt? 'Department of Natural Resources and Ecosystem Management, University of New England, Armidale, Ne^v South Wales 2351, Australia tBiological Dynamics of Forest Fragments Project, National Institute for Amazonian Research (INPA), C.P. 478, Manaus, AM 69011-970, Brazil ^iSchool of Renewable Natural Resources, 227 RNR Building, Louisiana State University, Baton Rouge, LA 70803-6202, U.S.A. Abstract: The impacts of potential linear barriers such as roads, highways, and power lines on rainforest fauna are poorly understood. In the central Brazilian Amazon, we compared the frequency of local movements (<300 m long) of understory birds within intact forest and across a 30-to 40-m-wide road over a 2-year period. Rainforest had regenerated along some road verges, to the extent that a nearly complete canopy was formed in some areas, so we also assessed whether this facilitated bird movement. Movement data were determined from 1212 recaptures of 3681 netted birds at six study sites. The road significantly inhibited total bird movement across roads at five of the six sites. Bird foraging guilds varied in their responses to the road and different ages of regrowth. Movements of frugivorous and edge and gap species were not inhibited at any site, whereas most forest-dependent insectivores (mixed-species flocks, terrestrial species, and army-ant followers) had markedly inhibited road-crossing movements, except at sites with extensive regrowth. Solitary understory species were especially vulnerable, rarely crossing even roads overgrown by tall regrowth. For sensitive species, road-crossing movements were inhibited because individuals tended to avoid both edge-affected habitat near the road and the road clearing itself Our results suggest that even narrow roads with low traffic volumes can reduce local movements of many insectivorous birds in Amazonia. Key Words: Amazonian birds, birds and roads, roads and bird movements Efectos de los Claros de Caminos sobre Patrones de Movimiento de Aves de Sotobosque en la Amazonia Central Resumen: Los impactos potenciales de barreras lineales como caminos, carreteras y l?neas el?ctricas sobre la fauna del bosque lluvioso son poco conocidos. Durante 2 a?os comparamos la frecuencia de desplaza- mientos locales (<300 m-longitud) de aves de sotobosque en selva intacta y de un lado a otro de un camino de 30-40 m de ancho en el Amazonas Brasile?o central La selva se hab?a regenerado en algunos bordes del camino, al grado que en algunas ?reas se hab?a formado un dosel casi completo, as? que tambi?n evalu- amos si esto facilitaba el desplazamiento de aves. Los datos de movimientos fueron determinados a partir de 1212 recapturas de 3681 aves en seis sitios de estudio. El camino significativamente inhibi? el desplaza- miento total de aves en cinco de los seis sitios. Los gremios de forrajeo variaron en su respuesta al camino y las diferentes edades de recrecimiento. Los desplazamientos de especies frug?voras y de borde y claros no fueron inhibidos en sitio alguno, mientras que los desplazamientos de la mayor?a de especies insect?voras dependientes de selva (parvadas de especies mixtas, especies terrestres y seguidoras de hormigas arrieras) fueron marcadamente inhibidos, excepto en los sitios con recrecimiento extensivo. Las especies solitarias de sotobosque fueron especialmente vulnerables, raramente cruzaron caminos, a?n los que ten?an recrecimiento '^Current address: Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Republic of Panama, email laurances@tivoU.sledu Paper submitted June 18, 2002; revised manuscript accepted November 4, 2003- 1099 Conservation Biology, Pages 1099-1109 Volume 18, No. 4, August 2004 1100 Road Effects on Bird Movements laurance et al. alto. Los desplazamientos especies sensibles de un lado a otro de caminos fueron inhibidos debido a que los individuos tendieron a evitar tanto al habitat afectado por borde cerca del camino como el claro del camino mismo. Nuestros resultados sugieren que aun los caminos angostos con bajos vol?menes de tr?fico pueden reducir los desplazamientos locales de muchas especies insect?voras en la Amazonia. Palabras Clave: aves amaz?nicas, aves y caminos, caminos y desplazamiento de aves Introduction The Amazon basin supports over half the ^vorld's remain- ing tropical rainforest (Whitmore 1997). Unfortunately, it also has the highest absolute rate of deforestation: 2-4 mUlion ha are being cleared each year, with a compara- ble area degraded by logging, forest fragmentation, and invasive ground fires (Nepstad et al. 1999; Laurance et al. 2001?; Cochrane & Laurance 2002). Moreover, devel- opment activities in Brazilian Amazonia are likely to ac- celerate during the next decade, with over $40 billion in planned infrastructure investments, including extensive linear clearings from new highw^ays, roads, pow^er lines, gas Unes, and river-channelization projects (Laurance et al. 2001&). The proliferation of human-made clearings may have important impacts on ^vildlife populations. To date, most fragmentation research has concentrated on large-scale clearings as barriers to w^ildlife movement, and it is only recently that narro^^ linear clearings, like those associated w^ith roads and pow^er lines, have been sho-wn to have sub- stantial effects on some species (revie^vs in Goosem 1997; Forman & Alexander 1998; Trombulak & Frissell 2000). In forests, linear clearings can cause important edge ef- fects (WiUiams-Linera 1990), promote invasion of exotic or generalist species (Goosem 1997), and increase human disturbances and hunting (W F. Laurance 2001; Kerley et al. 2002). For sensitive fauna, linear clearings may create partial or even complete barriers, impeding natural move- ments and gene flow and potentially fragmenting local populations (Bennett 1991; Forman & Alexander 1998). Most studies of linear clearings have focused on deter- mining their effects on the distribution and abundance of w^ildlife species in the adjacent habitat or on traffic- induced mortality. The influence of linear clearings on animal movements has been assessed less frequently, most often for small and medium-sized mammals (reviewed in Goosem 1997). Using baits, translocation, and playback of recorded calls, some researchers have demonstrated that w^ildlife species can be induced to cross a linear bar- rier (Burnett 1992; Desrochers &Hannon 1997; Goosem & Marsh 1997; Develey & Stouffer 2001). Although these are useful attempts to assess the movement capacity of species, they do not examine the patterns of passive (and presumably more natural) movements or home range boundaries across linear barriers. Understory rainforest birds are an ideal group for as- sessing the potential environmental effects of roads. They are highly diverse, readily sampled, and strongly associ- ated w^ith forest habitat. They have also been observed to decline or disappear in habitats that are disturbed by frag- mentation (Stouffer & Bierregaard 1995), logging (Thiol- lay 1992), and artificial forest edges (Canaday 1996; Re- strepo & Gomez 1998). Furthermore, they are generally considered poor colonists w^ith limited dispersal abilities (Lovejoy et al. 1986; Sieving & Karr 1997). For example, immigration by understory birds to central Amazonian forest fragments w^as sharply reduced by clearings as nar- ro^v as 70-100 m (Harper 1989; Stouffer & Bierregaard 1995). Some understory birds appear so specialized for dark forest conditions that they have been observed avoid- ing even tree-fall gaps and forest regrowth (ThioUay 1992; Borges & Stouffer 1999). We assessed the local movement patterns of Amazo- nian understory birds across roads and w^ithin adjoining forest over a 2-year period. By studying six road sites ^vith varying levels of forest regeneration w^e w^ere able to ad- dress several questions: Do roads affect the movement patterns of understory birds? Which foraging guilds are most affected? Does the presence of forest regro^vth along road verges facilitate road crossing by these birds? Methods Study Area This study w^as undertaken as part of the Biological Dy- namics of Forest Fragments Project (BDFFP), w^hich en- compasses a 1000-km^ study area located 80 km north of Manaus, Brazil (60?00'W, 2?20'S; for detailed descriptions, see Lovejoy et al. 1986; Laurance et al. 2002). The area supports nonflooded, low^land tropical rainforest, ^vhich averages 28-35 m in height and has emergent trees reach- ing 55 m. The understory is relatively open and domi- nated by stemless palms. Soils are yello^v latosols and are nutrient-poor and highly acidic. Rainfall averages about 2600 mm/year and is seasonal, w^ith a wet season from January to April and a dry season from June to Septem- ber The BDFFP study area is spanned along its east-w^est axis by a 40-km-long dirt road, which w^as bulldozed through Conservation Biology Volume 18, No. 4, August 2004 Lauramx et al. Road Effects on Bird Movements 1101 the forest in the late 1970s to provide access to local farms and cattle ranches. Originally, a clearing 30-40 m wide ^vas created, which included a 4- to 6-m-w^ide road surface and verges along either side. Regrowth vegeta- tion has regenerated along the road verge and varies in height and complexity. We examined three different lev- els of forest regeneration: (1) cleared-road sites supported a small amount of regro^vth (<3 m in height) that ex- tended about 7 m from the primary forest borders on each side, resulting in a road-clearing w^idth of about 25 m; (2) intermediate-regrowth sites had regrowth 3-8 m in height that extended about 14 m into each road verge, leaving a clearing of about 12 m wide; (3) tall-regrowth sites sup- ported regro^vth 15-20 m in height that extended in a noncontinuous canopy across the entire road clearing, w^ith some canopy gaps 2-3 m wide. All sites w^ere sep- arated by >500 m. Road traffic was restricted to autho- rized vehicles and averaged 6-10 passes per ^veek. In all six study sites, intact forest (unlogged and unfragmented) extended for >20 km on each side of the road. Sampling Methods At each of the six study sites, w^e assessed local bird move- ments (<300 m long) ^vithin a large study plot (4.5 ha). Plots w^ere centered on the road and consisted of six 150- m transects oriented parallel to the road and 60 m apart (Fig. 1). By comparing movement frequencies of birds betw^een adjacent net lines ^vithin the forest versus those across the road, w^e could quantitatively assess the effects of the road on bird movement (Fig. 1). We sampled each site five times, each time for 4 consec- utive days. Each day ^ve opened three neighboring mist nets on each line and afterward moved the nets to new adjoining locations along the line. Standard-size mist nets (36-mm mesh, 2 x 12 m) w^ere opened at 0600 hours for 8 hours on the first day and 6 hours on following days, except w^hen heavy rain required nets to be closed early. Although the total transect length netted w^as 150 m, only 36 m was netted per day on each transect. We sampled from November 1997 to November 1999, with an inter- val of 1.5-8 months betw^een samples (although sampling intervals varied somew^hat among sites, all sites w^ere sam- pled with similar frequency in the dry and w^et seasons). Nets were checked every 1-1.5 hrs, and captured birds w^ere identified, w^eighed, measured, and given unique numbered bands. We determined bird age w^here possi- ble based on plumage, eye color, bUl length, and presence of gape flanges. Vegetation Structure along Roadsides Profiles of roadside vegetation w^ere prepared for each study site to illustrate structural differences in vegetation among sites. Profiles were 60 m long and w^ere perpendic- ular to the road, ^vith the midpoint of the profile located at the road center They w^ere situated at the center of CT^ f?\ RCM WFM T1 T2 T3 T4 T5 T6 CD C ?|_ ro o ?o ro o Day 1 130 70 10 10 70 130 Distance from border (m) Distance from t>order (m) Figure 1. Design of mist-netting plots established at each study site. All net lines were separated by 60 m. Arrows at the top of the figure illustrate the types of movement considered in the analysis: road-crossing movements (RCM) and within-forest movements (WFM). each netting plot. Every canopy tree and sapling that in- tercepted the linear transect w^as mapped, including its position along the transect, canopy size, and height. We also recorded the height and position of all smaller plants (>1.5 m) w^ithin a 2-m-w^ide band along the transect. In addition to the vegetation profiles, quantitative studies revealed significant changes in several forest structural variables -within 10-15 m of road edges (reduced canopy height and cover, increased subcanopy cover and logs) relative to forest interiors (130 m from edge; S. G. W Lau- rance 2001). Understory Birds and Their Foraging Guilds The central Amazon rainforest supports a highly diverse avifauna community w^ith approximately 400 species in the vicinity of the BDFFP study area (Stotz & Bierregaard 1989; Cohn-Haft et al. 1997). Insectivorous species, par- ticularly antbirds (Formicariidae), dominate the under- story community (Bierregaard 1990). We placed the most common insectivorous species into guilds, according to their foraging behavior and habitat use (Stouffer & Bier- regaard 1995; Cohn-Haft et al. 1997). Army-ant follow- ers are the most frequently captured guild, comprised of Conservation Biology Volume 18, No. 4, August 2004 1102 Road Effects on Bird Movements laurance et al. three obligate species that feed almost exclusively on in- sects fleeing army-ant swarms (Willis & Oniki 1978). An- other common guild is mixed-species flocks, w^hich are assemblages of 10-20 species pairs that forage and travel together (Pow^ell 1989)- More than a dozen species can occur w^ithin a flock at any one time, although the degree of flocking propensity varies among species (JuUien & ThioUay 1998). Many other understory species maintain a territory as a pair or family, yet occasionally join ant fol- lo^vers and mixed-species flocks w^hen they pass through their territories (Harper 1989; Po^vell 1989). Among these nonflocking species, w^e divided birds on the basis of their habitat use into three guilds: terrestrial species that for- age in the leaf litter and rarely fly as they forage, edge and gap species that prefer forest ecotones and treefaU gaps, and solitary understory species that glean foliage or sally to capture insects in the understory. Understory frugivores are also commonly captured in mist nets and were examined in this study. Data Analysis Two potentially distinctive behavioral mechanisms may collectively determine the frequency wth which w^ildlife ^viU traverse roads or clearings. The first mechanism is edge avoidance. Some species decline in abundance near forest edges (Qu?ntela 1985; Restrepo & Gomez 1998; S. G. W. Laurance 2004) and may rarely cross road clearings simply because they are uncommon near roads. The sec- ond mechanism is gap avoidance, w^hereby species avoid forest clearings (Greenberg 1989). To our know^ledge, the relative importance of these two mechanisms, and the de- gree to w^hich they are correlated w^ithin guilds, has not been assessed previously. The net effect of road clearing on bird movements re- sults from the combined effects of edge and gap avoid- ance. We evaluated the net effect of the road on bird movements w^ith mark-recapture data by comparing loca- tions between consecutive captures. Our null hypothesis was that if the road had no net effect on bird movement, then the frequency of recorded movements betw^een net lines on opposite sides of the road should not differ sig- nificantly from that of similar lines in nearby intact for- est. We considered that the frequency of bird movements between adjacent net lines, w^hich w^ere 60 m apart, rep- resented 60-m movements irrespective of diagonal dis- tances. The same applied to movements betw^een lines that w^ere 120 m apart. Recorded bird movements in- cluded those that occurred w^ithin a single 4-day netting session and those occurring between different trapping sessions. Few movements w^ere recorded between differ- ent study sites (4.1% of total captures), and these w^ere excluded from the analysis. In our study design (Fig. 1), if the road had no net effect on bird movement, there should have been four times as many 60-m movements wthin the forest (between adja- cent net lines Tl and T2; T2 and T3; T4 and T5; and T5 and T6) as across the road (between adjacent lines T3 and T4). In addition, there should have been equal numbers of 120-m movements w^ithin the forest (betw^een lines Tl and T3; and T4 and T6) as across the road (betw^een lines T2 and T4; and T3 and T5). Thus, the total number of expected road-crossing movements for any group of birds was 20% of their recorded 60-m movements, plus 50% of their recorded 120-m movements. For example, if a species had 20 movements of 60 m and 4 movements of 120 m, then their expected road-crossing movements would be 6. The 60- and 120-m movements w^ere pooled because the latter w^ere relatively infrequent (<29% of all movements) and hence could not be assessed separately, and because pooling the data increased sample sizes for rare species and guilds. The net effect of the road on bird movement w^as assessed at each site with G tests for goodness of fit, with WiUiams's correction for sample size (Sokal & Rohlf 1995). Data at replicate sites w^ere pooled according to regro^vth type w^hen bird communities at those sites re- sponded similarly to the road (Bailey 1995), meaning that if both sites show^ed either significant or nonsignificant results they w^ere pooled. We assessed net road effects for both the entire bird community and several distinct feed- ing guilds (Appendix 1). A simple index of road avoidance was generated to illustrate bird responses to road clear- ings: net road avoidance = [1 ? (observed road cross- ings/expected road crossings)] x 100. In addition, w^e quantified edge and gap avoidance for each feeding guild (limited sample sizes precluded these analyses for individual species). An index of edge avoid- ance w^as calculated for each guild by dividing its mean abundance (no. individuals/1000 net hours) on net lines farthest from the road (130 m from the road verge) by its abundance on net lines nearest the road (10-15 m from the verge). We tested gap avoidance for each guild by con- sidering only bird movements that originated on net lines nearest the road (thereby eliminating any effect of edge avoidance) and then comparing the frequency of w^ithin- forest versus road-crossing movements, w^ith G tests for goodness of fit w^ith WiUiams's correction for sample size. For this analysis, if the birds exhibited no gap avoidance, then 50% of all 60- and 120-m movements should be road- crossing movements, and 50% should be w^ithin-forest movements. We summarized these data for each guild with the following index: gap avoidance = (observed road crossings ? expected road crossings)/expected road crossings. For these analyses, data for tall-regrowth sites w^ere excluded to reduce the potentially confounding ef- fects of dense regro^^th on edge and gap avoidance. Movements of longer distances (180-300 m) w^ithin the same study plot, which inevitably involved road cross- ings, occurred relatively infrequently and could not be compared statistically to the frequency of w^ithin-forest movements because the maximum possible length of Conservation Biology Volume 18, No. 4, August 2004 Lauramx et al. Road Effects on Bird Movements 1103 within-forest movements was 120 m (Fig. 1). We did as- sess these data qualitatively, how^ever, and contrasted their relative frequency among different bird guilds and among study sites w^ith differing levels of regrowth ^vith a tw^o- way analysis of variance (ANOVA). Results A total of 3681 captures w^as recorded from 13,176 mist- net hours. Bird recaptures w^ere about one-third (1212) of aU captures, of w^hich 810 w^ere recorded bird move- ments (i.e., birds captured in one net line then recap- tured in a different line). Of 116 species w^e caught, w^e recaptured ? at least once and recorded movements of 56 (Appendix 1). The 10 most commonly recaptured species were Hylophylax poeciltnota, Glyphorynchus spirurus, Gymnopithys rufigula, Thaninomanes arde- siacus, Pithys albifrons, Myrmotherula gutturalts, Xiphorhynchus pardalotus, Percnostola ruftfrons, and Turdus albtcolUs. Total Bird Movements Roads sharply reduced the net rate of bird movement. For five of the six study sites, significantly fe^ver road crossings occurred than w^ere expected by chance (Fig. 2a). Road crossings for all birds w^ere reduced at cleared sites (Gadj = 12.45,/? = 0.0004), intermediate-regrowth sites (Gadj = 4.77, p = 0.029), and tall-regrowth site 2 (Gad] = 7.75, p = 0.006), but not at tall-regrowth site 1 (G^di = 0.15,/7 = 0.70; df = 1 in aU cases), which had the tallest and most continuous canopy cover of any site. At each site, on average more than tw^ice as many bird species w^ere recorded moving w^ithin the forest (x ? SD = 26.50 ? 3.02 species) as across roads (11.83 ? 2.12 species). For all sites combined, 50 species w^ere recorded moving ^vithin the forest, whereas only 31 species crossed roads (Appendix 1). Road-crossing species w^ere a lim- ited subset of those moving w^ithin the forest, except for Corythopis torquata and Hylexetastes perrotii, both of w^hich are forest birds ^vhose lack of recorded w^ithin- forest movements most likely resulted from lo^v abun- dances rather than a preference for edge habitats. Juvenile Birds We analyzed net movements of juvenile birds separately (Fig. 2b) because they are more likely to be transient and might cross roads in search of vacant territories. Ju- veniles exhibited a significant reduction in movements at cleared sites (Gadj = 4.31, p = 0.038), but not at ei- ther intermediate-regrowth (Gadj = 0.95, p = 0.33) or taU-regrowth (Gadj = 0.71, p = 0.40) sites. Foraging Guilds Three species of army-ant foUow^ers in our study area are considered obligate (or "professional") ant-foUowng birds: Pithys albifrons, Gymnopithys rufigula, and Den- drocincla nierula. AU w^ere recorded moving within the forest at all six sites, but at five of the six sites only one or tw^o species w^ere recorded crossing the road (Fig. 2c). The net rate of movement across the road was signifi- cantly reduced at cleared sites (Gadj = 7.19, p = 0.007), intermediate-regrowth sites (Gadj = 4.06,p = 0.044), and taU-regrowth site 2 (Gadj = 7.01,p = 0.008) but not at tall- regrowth site 1 (Gadj = 0.39,/? = 0.53). The White-plumed Antbird (Pithys albifrons') was the most frequently cap- tured bird in this guild for both w^ithin-forest and across- road movements. Nine solitary understory species had recorded move- ments between net lines (Fig. 2d). Overall, this guild ex- hibited the highest degree of sensitivity: net road-crossing movements w^ere sharply reduced at all six study sites (cleared: Gadj = 5.08,/? = 0.024; intermediate regrowth: Gadj = 11.04,/? = 0.0008; tall regrowth: Gadj = 9.43,/? = 0.002). Fifteen species that regularly forage in mixed-species flocks had recorded movements (Fig. 2e). Net road- crossing movements w^ere significantly less frequent than expected at cleared sites (Gadj = 8.84,/? = 0.003) and at tall-regrowth site 2 (Gadj = 7.69, /? = 0.006). How^ever, road-crossing movements were as frequent as expected at intermediate-regrowth sites (Gadj = 145,/? = 0.23) and nearly more frequent than expected at tall-regro^^th site 1 (Gadj = 3.80, /? = 0.051), where flocks often for- aged in the mature regrowth spanning the road (S. G. W Laurance, personal observation). Almost tw^o-thirds of the road crossings in this guild w^ere made by just tw^o species, Xiphorhynchus pardalotus and Glyphorynchus spirurus (Appendix 1). Movements w^ere recorded for seven terrestrial species, but these birds w^ere uncommon in mist-net captures be- cause they usually walk rather than fly through the for- est. Despite the small number of recorded movements (n = 20), significantly fewer movements than expected occurred at cleared sites (Gadj = 4.08,/? = 0.043) but not at sites of intermediate regrowth (Gadj = 0.05,/? = 0.82) and tall regrowth (Gadj = 0.30,/? = 0.59) (Fig. 2f). Six insectivore species that commonly forage on forest edges or gaps had recorded movements (Fig. 2g). For this guUd, there w^as no net inhibitory effect of the road at cleared sites (Gadj = 1-25,/? = 0.26) and tall-regrowth sites (Gadj = 111,/? = 0.29), and road crossings w^ere actually more frequent than expected at intermediate-regrowth sites (Gadj = 4.94,/? = 0.026). For seven frugivore species (Fig. 2h), no significant net effects of the road w^ere detected at any study site (cleared sites: Gadj = 0.44,p = 0.51; intermediate-regrowth sites: Gadj = 0.08, /? = 0.78; tall-regrowth sites: Gadj = 0.60, p = 0.40). The most abundant member of this guUd, the White-crow^ned Manakin (Piprapipra), accounted for about 70% of road crossings (Appendix 1). Two other fru- givores, Schiffornis turdinus and Turdus albicolUs, may Conservation Biology Volume 18, No. 4, August 2004 1104 Road Effects on Bird Movements laurance et al. X u Q Z u u z -< Q O > < Q < O 06 100 50 -SO e) Mixed-species flocks Tan Fc^rowth I TaU rtgnmtkl 100 80 60 40 u 20 a z 0 M u -20 1 -40 a 20 o > 0 < % -20 o C? -40 - m m 0 Terrestrial birds - y////y w> - h) Frugivores Till rc{[row1h Figure 2. Road avoidance by insectivorous birds in central Amazonia in response to differing levels of roadside vegetation. Positive road-avoidance values indicate the road reduced bird movements relative to rates inside the forest, whereas negative values indicate an opposite trend (significance levels: *p < 0.05; **p < 0.01; ***p < 0.001). be more sensitive to roads than this guUd-level analysis suggests. For these two species, there w^ere 26 within- forest movements and only a single road-crossing move- ment. Edge and Gap Avoidance Edge avoidance varied greatly among the six guilds. As indicated by edge-avoidance values of < 1, edge and gap species (0.45) and frugivores (0.91) increased in abun- dance near the road, w^hereas army-ant foUow^ers (2.00), solitary species (2.41), mixed-species flocks (1.79), and terrestrial species (2.52) all declined near the road (Table 1). Gap avoidance also varied among guilds, with positive values indicating gap avoidance and negative values indi- cating the opposite trend. Despite limited sample sizes in this particular analysis (Table 1), army-ant foUow^ers (G^dj = 8.0,/7 = 0.005) and solitary species (G^?) = 3.85, p = 0.05) exhibited significant gap avoidance, and results w^ere nearly significant for mixed-species flocks (Gadj = 3.60,/? = 0.058). Terrestrial species could not be tested because of inadequate sample sizes, whereas edge and gap species (Gad] = 0.35, p = 0.55) and frugivores (Gad] = 0.3l,p = 0.58) showed no significant gap avoidance. Conservation Biology Volume 18, No. 4, August 2004 Lauramx et al. Road Effects on Bird Movements 1105 Table 1. Data used to derive indices of edge and gap avoidance for each guild in this study." Mean abundance' Edge avoidance Observed movements^ Gap Guild edges interiors within forest across road avoidance Ant followers 10.84 21.72 2.00 22 7 0.52 Solitary species 5.86 14.11 2.41 8 2 0.60 Mixed-species flocks 14.94 26.77 1.79 30 17 0.28 Terrestrial species 1.93 4.88 2.52 1 1 ? Edge/gap species 11.32 5.08 0.45 11 14 -0.12 Frugivores 13.61 12.44 0.91 16 13 0.10 "To avoid biasing results, we excluded tall-regrowth sites from these analyses. ''Number of individuals per 1000 net hours. '^Includes only movements that originated on net lines adjoining the road. For most guilds, the indices of edge and gap avoidance were strongly and positively associated (Fig. 3). Except for terrestrial species, which ^vere captured too infrequently near edges to allo^v reliable estimates of gap avoidance (Table 1), the remaining five guilds exhibited a highly significant correlation between edge and gap avoidance (r = 0.976, df = 3,^ = 0.004; Pearson correlation). This pattern suggests that, at least among the avian guilds ex- amined in our study, edge and gap avoidance are corre- lated traits and both likely contribute to the net reduction of road-crossing movements in sensitive species. Long-Distance Movements In addition to localized movements of 60-120 m, some birds w^ere recorded moving longer distances (180-300 m) between net lines at the same study plot, which in- evitably involved crossing the road (Table 2). Although 0.6- Solitary spp. ? 0.5- Ant-followers ? X ? 0.4- r=0.977,/;==0.004 / cs 1 -g 0.3- ?? cs 0.2- / y Mixed-spp. flocks & - Frugivores X O 0.1- ? / 0- / -0.1- ^ Edge/gap spp. 1 1 1 ' '1 "?? 1 0.5 1 1.5 Edge avoidance 2.5 Figure 3- Positive association between indices of edge avoidance and gap avoidance in five guilds of Amazonian understory birds. no experimental controls are available to quantify the fre- quency of such movements in intact forest, the relative frequency of long-distance movements (as a proportion of all recorded movements) can be compared among each guUd and study area (Table 2). This analysis, using a tw^o- w^ay ANOX'A w^ith arcsine-transformed data, revealed that the guilds varied significantly in their frequency of long- distance movements (i's.is = 317,/? = 0.032). On av- erage, army-ant foUow^ers exhibited a significantly higher frequency (p < 0.05) of long-distance movements (28% of all movements) than did either edge and gap species (3%) or solitary understory species (5%) (Tukey's post hoc tests). The remaining guilds and juvenile birds had in- termediate frequencies of long-distance movements (13- 15%). There was no significant effect of regrowth type on the frequency of long-distance movements (7^2, is = 0.95, /? = 0.41) or any interaction betw^een guild and regro^^th type (i'lo.is = 0.45,^ = 0.90). Thus, for highly mobile species like army-ant foUow^ers, the potential barrier ef- fects of roads may be less severe than for more sedentary, forest-dependent species like solitary understory insecti- Discussion Our results suggest that even a relatively narro^v, un- paved road ^vith little traffic significantly reduces the road-crossing movements of many Amazonian understory birds. This reduction in movements apparently has tw^o distinct but interrelated causes: edge avoidance and gap avoidance. Sensitive species exhibited edge avoidance, tending to avoid the general vicinity of forest edges near roads, and they exhibited gap avoidance, rarely traversing the road clearing itself. At least at a guild level, these tw^o behaviors were strongly and positively correlated (Fig. 3), and both contributed to the observed impact of the road on bird movements. To our know^ledge, ours is the first study to distinguish the effects of these two distinctive mechanisms that can lead to reduced w^ildlife movements across roads. Conservation Biology Volume 18, No. 4, August 2004 1106 Road Effects on Bird Movements laurance et al. Table 2. Percentage (and number) of long-distance movements (180-300 m) recorded for eacb avian guild and study site. Cleared sites Medium regrowth site 1 site 2 Tall regrowth Guild site 1 site 2 site 1 site 2 Mean (sum) Ant followers 32(7) 41(11) 22 (8) 25 (14) 29 (20) 23(8) 28 (68) Solitary species 7 (1) 0 (0) 0 (0) 5 (1) 19(3) 0(0) 5(5) Mixed-species flocks 3 (1) 16 (6) 10 (4) 17 (7) 18(11) 27 (10) 15 (39) Terrestrial species 0 (0) 14 (1) 50 (1) 0 (0) 50(1) 0(0) 13(3) Edge/gap species 0 (0) 0 (0) 5 (1) 10 (1) 0(0) 0(0) 3(2) Frugivores 6 (1) 0 (0) 11(2) 26(6) 15(2) 11(2) 13(13) Juveniles 5 (1) 22 (4) 12 (4) 27 (7) 5(1) 9(2) 14 (19) Total birds 10 (10) 17 (18) 12 (16) 19 (29) 22(37) 15 (20) 16 (130) Mechanisms Promoting Road Avoidance Why do many Amazonian understory birds avoid forest edges and clearings? First, many species may have had lit- tle reason to traverse clearings in their evolutionary his- tory so avoidance of such areas is probably an innate response (e.g., Greenberg 1989). Interestingly, species that readily traverse road clearings, such as the mixed- species flock members Glyphorynchus spirurus and Xiphorhynchus pardalotus, have persisted in small (1- 10 ha) forest fragments in our study area, whereas other members of this guild, w^hich rarely use road clearings, have declined significantly or disappeared (Stouffer & Bierregaard 1995; S. G. W Laurance 2001). Second, abrupt forest ecotones near roads and clear- ings are subject to diverse edge effects that alter forest structure, light levels, thermal regimes, floristic compo- sition, and invertebrate abundance (e.g., Lovejoy et al. 1986; Kapos 1989; Didham 1997; Laurance et al. 1998, 2002; S. G. W Laurance 2001). Such environments prob- ably seem harsh or unfamiliar to bird species adapted to the dark, humid conditions of forest interiors. Edge avoid- ance might also occur because diurnal predators such as White Haw^ks (Leucopternis albicollis) and Bat Falcons (Falco rufigularis) tend to hunt along forest edges (R. O. Bierregaard, personal communication). Territorial behav- ior by invasive edge or generaUst species could potentially cause edge avoidance, but incursions of non rainforest birds (e.g.. Troglodytes aedon, Ramphocelus carbo) are rare in our study area except in large clearings (Stouffer & Bierregaard 1995; Stouffer & Borges 2001). Third, territorial animals such as birds are known to align their territories or home ranges along sharp habitat discontinuities such as road margins (review^ed in Bennett 1991; Goosem 1997; Forman & Alexander 1998). Nor- mal territorial defense by resident birds would therefore tend to reduce road-crossing movements. For guilds such as mixed-species flocks, how^ever, regeneration of tall re- growth along road verges "softened" the forest edge to the degree that the entire road was incorporated into the birds' home range (Develey & Stoviffer 2001). Other species, such as certain edge and gap insectivores, ap- peared to establish home ranges that spanned the road and adjoining forest on either side, especially w^here there was limited regrowth along road verges (S. G. W Laurance 2001). Finally, some w^ildUfe species avoid the vicinity of roads because of human or traffic-related disturbances (e.g., Rost 8L Bailey 1979; van der Zande et al. 1980; Goosem 1997). Songbirds, for example, appear sensitive to even low levels of noise (Forman & Alexander 1998). Such ef- fects are unlikely in this study, how^ever, because traffic volume was low (6-10 vehicle passes/w^eek) and because our study area is protected from hunters, loggers, and other human disturbances. There was no observed road mortality of birds during this study. Effects of the Road and Adjoining Regrowth The narrow road w^e studied sharply reduced overall bird movements across the road, but the net effect of the road varied greatly among birds in different foraging guUds. Two guilds, frugivores and edge- and gap-favoring insec- tivores, were little affected by the road, regardless of the clearing width and amount of adjoining forest regro^vth. Two other guilds, mixed-species flocks and terrestrial species, exhibited significantly reduced movement across cleared roads w^ith little regrowth along their verges. Finally, the tw^o remaining guilds, solitary understory species and army-ant foUow^ers, appeared to be strongly affected by road clearings, w^ith sharply diminished move- ments across both cleared roads and those w^ith substan- tial regrov^th along their verges. These conclusions, how^ever, must be tempered by the fact that our study focused on short-distance move- ments (5120 m) across roads and in forest. Some longer- distance movements (180-300 m) across roads also oc- curred, especially by army-ant foUow^ers (Table 2), prob- ably because they often forage w^idely w^ith fev^ fixed ter- ritorial boundaries as they track marauding swarms of army ants (Willis & Oniki 1978; Harper 1989). This sug- gests that army-ant followers in particular, and highly mo- bile species in general, may be less vulnerable to road clearings than are sedentary, territorial species, especially when the latter are strongly dependent on forests. No- tably, experimental studies of the ant-foUowng species Conservation Biology Volume 18, No. 4, August 2004 Lauramx et al. Road Effects on Bird Movements 1107 have shown that birds translocated into forest fragments w?l cross clearings 100 to 320 m w^ide to return to pri- mary forest (Harper 1989), although clearings of only 100 m preclude their movements under normal circum- stances (Bierregaard & Lovejoy 1989; Stouffer & Bierre- gaard 1995). Our results are in accord w^ith those of other studies suggesting that many understory insectivorous birds are highly sensitive to habitat alteration (ThioUay 1992; Stouf- fer & Bierregaard 1995; Canaday 1996; Sieving & Karr 1997; Restrepo & Gomez 1998; Stratford & Stouffer 1999). Our findings may be conservative for guilds such as terres- trial species that are relatively rare and difficult to capture in mist nets. Most terrestrial species are highly vulnerable to forest fragmentation (Stratford & Stouffer 1999) and may have exhibited more significant, negative responses to roads had our sample sizes been larger For certain guilds, the height and density of forest regrowth along road verges had a strong effect on bird movement. This was especially evident for mixed-spe- cies flocks, which had fe^v short-distance movements across cleared roads but appeared little affected at sites w^ith w^ell-developed regrowth spanning the road. At taU- regrowth site 1, w^hich had the most mature regrowth of any site, at least one mixed-species flock incorporated the road into its home range, as the birds w^ere often ob- served foraging in trees adjoining and above the road. Sim- ilarly, edge- and gap-favoring species appeared to favor road sites w^ith intermediate regrowth, actually travers- ing them more frequently than they did in intact forest. Solitary insectivores rarely crossed roads, how^ever, even those with considerable regrowth along their verges. Although clear differences w^ere evident among the dif- ferent bird guilds, species w^ithin the same guild occasion- ally responded very differently to roads. As a group, for example, understory frugivores appeared little affected by roads, and some species fed on fruits of w^eedy shrubs (Family Melastomataceae) that proliferated along road verges. Ho-wever, two frugivore species, Schiffornis tur- dinus and Turdus albicollis, appeared to be strongly in- hibited by the road (Table 1), and both are also far more vulnerable than other frugivores to forest fragmentation (Bierregaard & Stouffer 1997). even more vulnerable. For such species, it is not incon- ceivable that major roads could subdivide populations, in- creasing their vulnerability to random demographic and genetic effects and increasing the probability of local ex- tinction (Wilcox & Murphy 1985; Gilpin & Soul? 1986). Among the avian guilds w^e studied, the movement- inhibiting effects of the road appeared greatest for more sedentary, forest-specialist species. Solitary understory insectivores w^ere most vulnerable, foUow^ed by mixed- species flocks, ant-foUow^ing birds, and terrestrial species. These species often decline in abundance near forest edges, avoid clearings (S. G. W Laurance 2001), and are highly vulnerable to forest fragmentation (Bierregaard & Lovejoy 1989; Stouffer & Bierregaard 1995; Stratford & Stouffer 1999). Notably, these same guilds are most strongly affected by selective logging operations, w^hich create networks of roads and small clearings in forests (Thiollay 1992). Regrowth along road verges markedly affected road- crossing movements by forest birds. Our road site w^ith the tallest regrowth, w^hich formed a nearly continuous canopy over the road, was readily used by mixed-species flocks and ant foUow^ers, guilds that rarely crossed road sites with little regro^^th. In nature reserves, resource managers can help minimize the impacts of roads on sen- sitive w^ildUfe by encouraging forest regeneration along road verges and establishing continuous canopy cover over the road surface. Roads, power lines, and other lin- ear clearings in reserves should be minimized and the w^idths of clearings limited (i.e., to no more than 15-20m) so that nearly continuous canopy cover is maintained above the clearing. The dramatic gro^vth of logging operations, forest- colonization projects, and new infrastructure develop- ments in the Amazon (Laurance et al. 2001&) suggests that the extent and number of linear clearings w^iU in- crease sharply in the future. By greatly increasing physical access to frontier areas, such projects could sharply accel- erate rates of forest loss and fragmentation (Laurance et al. 2001??, 2001&). Efforts are needed to plan and actively manage the design of roads and other linear clearings in the Amazon to reduce their deleterious effects on forest- dependent wUdllfe. Implications for Conservation Our results have important implications for forest conser- vation. First, the narro^v dirt road w^e studied had surpris- ingly large effects on some species and guilds of under- story insectivorous birds. Indeed, our results are probably a best-case scenario because our study area is protected from hunters, loggers, and heavy vehicle use, which w^ould increase disturbances to w^ildlife. If the movements of mobile organisms such as birds are being inhibited by this narro^v road, then less mobile species, such as ar- boreal vertebrates and sedentary invertebrates, might be Acknowledgments We sincerely thank E Jarman for supervising the doctoral project of S.G.WL; O. de Souza Pereira, C. Strong, andj. Long for field assistance; and R. Bierregaard, M. Cohn- Haft, C. Marantz, D. Robinson, T Donovan, and three anonymous referees for many helpful comments on the manuscript. Financial support w^as provided by the Bio- logical Dynamics of Forest Fragments Project, Australian Postgraduate Aw^ards, the University of New England, and the National Geographic Society. This is publication 408 Conservation Biology Volume 18, No. 4, August 2004 1108 Road Effects on Bird Movements laurance et al. in the Biological Dynamics of Forest Fragments Project technical series. Literature Cited Bailey, N. T. J. 1995. Statistical methods in biology. 3rd edition. Cam- bridge University Press, Cambridge, United Kingdom. Bennett, A. F. 1991 ? Roads, roadsides and mldllfe conservation: a review?. Pages 99-117 in D. Saunders and R. Hobbs, editors. Nature conser- vation 2: the role of corridors. Surrey Beatty and Sons, Sydney. Bierregaard, R. O., Jr 1990. Species composition and trophic organiza- tion of the understory bird commtmity in a central Amazonian terra firme forest. Pages 217-236 in A. Gentry, editor. Four Neotropical rainforests. Yale University Press, New? Haven, Connecticut. Bierregaard, R. 0.,Jr, andT. E. Lovejoy 1989- Effects of forest fragmen- tation on Amazonian understory bird communities. Acta Amaz?nica 19:215-241. Bierregaard, R. O., Jr, and P. C. Stotiffer. 1997. Understory birds and dynamic habitat mosaics in Amazonian rainforest. Pages 138-155 in W. F Laurance and R. O. Bierregaard Jr, editors. Tropical forest remnants: ecology, management, and conservation of fragmented communities. University of Chicago Press, Chicago. Borges, S., and P C. Stotiffer. 1999. Bird communities in two types of anthropogenic successional vegetation in central Amazonia. Condor 101:529-536 Burnett, S. 1992. Effects of a rainforest road on movements of small mammals: mechanisms and implications. Wildlife Research 19: 95-104. Canaday, C. 1996. Loss of insectivorous birds along a gradient of htiman impact in Amazonia. Biological Conservation 77:63-71. Cochrane, M. A., and W. F Laurance. 2002. Fire as a large-scale edge effect in Amazonian forests. Journal of Tropical Ecology 18:311-325. Cohn-Haft, M., A. Whittaker, and P C. Stouffer. 1997. A new look at the "species-poor" central Amazon: the avifauna north of Manaus, Brazil. Ornithological Monographs 48:205-235. Desrochers, A., and S. J. Hannon. 1997. Gap crossing decisions by for- est songbirds during the post-fledging period. Conservation Biology 11:1204-1210. Develey, P., and P C. Stouffer. 2001. Effects of roads on movements of tmderstory birds in mixed-species flocks in central Amazonian Brazil. Conservation Biology 15:1416-1422. Didham, R. K. 1997. The influence of edge effects and forest fragmen- tation on leaf-Utter invertebrates in central Amazonia. Pages 55-70 in W. F Laurance and R. O. Bierregaard Jr, editors. Tropical forest remnants: ecology, management, and conservation of fragmented communities. University of Chicago Press, Chicago. Forman, R. T T, and L. E. Alexander 1998. Roads and their major ecolog- ical effects. Annual Review of Ecology and Systematics 29:207-231- Gilpin, M. E., and M. E. Sotil?. 1986. Minimum viable populations: pro- cesses of species extinction. Pages 19-34 in M. E. Soul?, editor Con- servation biology: the science of scarcity and diversity. Sinauer As- sociates, Sunderland, Massachusetts. Goosem, M. 1997. Internal fragmentation: the effects of roads, highways and pow^erline clearings on movement and mortality' of rainforest vertebrates. Pages 241-255 in W. F Laurance and R. O. Bierregaard Jr., editors. Tropical forest remnants: ecology, management, and con- servation of fragmented communities. University of Chicago Press, Chicago. Goosem, M., and H. Marsh. 1997. Fragmentation of a small mammal com- munity by a pow^erline corridor through tropical rainforest. Wudlife Research 24:613-629. Greenberg, R. 1989. Neophobia, aversion to open space, and ecological plasticity' in Song and Swamp Sparrows. Canadian Journal of Zoology 67:1194-1199. Harper, L. H. 1989. The persistence of ant-following birds in small Ama- zonian forest fragments. Acta Amaz?nica 19:249-263. Jul?en, M., and J. M. Thio?ay 1998. Multi-species territoriality and dy- namics of Neotropical forest tmderstorey bird flocks. Journal of An- imal Ecology 67:227-252. Kapos, V. 1989. Effects of isolation on the w^ater status of forest patches in the Brazilian Amazon. Journal of Tropical Ecology 5:173-185. Kerley, L. L., J. Goodrich, D. MiqueUe, E. Smirnov, H. Quigley, and M. Hornocker. 2002. Effects of roads and htunan disturbance on Amur tigers. Conservation Biology 16:97-108. Laurance, S. G. W. 2001. The effects of roads and their edges on the move- ment patterns and community' composition of tmderstorey rainfor- est birds in central Amazonia, Brazil. Ph.D. thesis. University of New England, Armidale, Australia. Laurance, S. G. W. 2004. Responses of understory rainforest birds to road edges in central Amazonia. Ecological Applications (in press). Laurance, W. F. 2001. Tropical logging and htmian invasions. Conserva- tion Biology 15:4-5. Laurance, W. E, A. Albernaz, and C. Da Costa. 2001?. Is deforestation accelerating in the Brazilian Amazon? Environmental Conservation 28:305-311. Laurance, W. F, M. A. Cochrane, S. Bergen, P M. Fearnside, P De- lam?nica, C. Barber, S. D'Angelo, and T. Fernandes. 2001&. The fu- ture of the Brazilian Amazon. Science 291:438-439. Laurance, W F, L. Y Ferreira, J. M. Rankin-deMerona, and S. G. Laurance. 1998. Rainforest fragmentation and the dynamics of Amazonian tree communities. Ecology 79:2032-2040. Laurance, W. E, T. E. Lovejoy, H. Vasconcelos, E. Bruna, R. Didham, P Stouffer, C. Gascon, R. O. Bierregaard Jr., S. G. Laurance, and E. Sampiao. 2002. Ecosystem decay of Amazonian forest fragments: a 22-year investigation. Conservation Biology 16:605-618. Lovejoy, T, et al. 1986. Edge and other effects of isolation on Amazon forest fragments. Pages 257-285 in M. Soul?, editor. Conservation biology: the science of scarcity and diversity. Sinauer Associates, Sunderland, Massachusetts. Nepstad, D. C, et al. 1999. Large-scale impoverishment of Amazonian forests by logging and fire. Nature 398:505-508. Powell, G. V. N. 1989. On the possible contribution of mixed species flocks to species richness in Neotropical avifaimas. Behavioral Ecol- ogy and Sociobiology 24:387-393- Qu?ntela, C. E. 1985. Forest fragmentation and differential use of natural and man-made edges by understory birds in central Amazonia. M.Sc. thesis. University of UUnois, Chicago. Restrepo, C, and N. Gomez. 1998. Responses of understory birds to anthropogenic edges in a Neotropical montane forest. Ecological Applications 8:170-183. Rost, G. R., and J. A. Bailey. 1979. Distribution of mule deer and elk in relation to roads. Journal of Wildlife Management 43:634-641. Sieving, K. E., and J. R. Karr. 1997. Avian extinction and persistence mechanisms in lowland Panama. Pages 156-170 in W F Laurance and R. O. Bierregaard Jr., editors. Tropical forest remnants: ecology, management, and conservation of fragmented communities. Univer- sity of Chicago Press, Chicago, Illinois. Sokal, R. F, and F H. J. Rohlf 1995. Biometry. 3rd edition.W H. Freeman, San Francisco. Stotz, D. F, and R. O. Bierregaard Jr 1989. The birds of the fazendas Porto Alegre, Dimona and Estelo north of Manaus, Brazil. Revista Brasileira de Biolog?a 49:861-872. Stouffer, P C, and R. O. Bierregaard Jr 1995. Use of Amazonian forest fragments by tmderstory insectivorous birds. Ecology 76: 2429-2443. Stouffer, P. C, and S. H. Borges. 2001. Conservation recommendations for tmderstory birds in Amazonian forest fragments and secondary areas. Pages 248-261 in R. O. Bierregaard Jr, C. Gascon, T E. Lovejoy, and R. Mesquita, editors. Lessons from Amazonia: ecology and con- servation of a fragmented forest. Yale University Press, New Haven, Connecticut. Stratford, J., and P. C. Stotiffer. 1999. Local extinctions of terrestrial insectivorous birds in a fragmented landscape near Manaus, Brazil. Conservation Biology 13:1416-1423. Conservation Biology Volume 18, No. 4, August 2004 Lauramx et al. Road Effects on Bird Movements 1109 ThioUay, J. M. 1992. Influence of selective logging on bird species di- versity in a Guianian rain forest. Conservation Biology 6:47-63. Trombulak, S. C, and C. A. Frisse?. 2000. Review of ecological effects of roads on terrestrial and aquatic communities. Conservation Biology 14:18-30. Whitmore, T. C. 1997. Tropical forest disturbance, disappearance, and species loss. Pages 3-12 in W. F. Laurance and R. O. Bierregaard Jr, editors. Tropical forest remnants: ecology, management, and con- servation of fragmented communities. University' of Chicago Press, Chicago. Wilcox, B. A., and D. D. Murphy. 1985. Conservation strategy: effects of fragmentation on extinction. The American Naturalist 125:879-887. Williams-Linera, G. 1990. Vegetation structure and environmental con- ditions of forest edges in Panama. Journal of Ecology 78:356-373. Willis, E. O., and Y. Oniki. 1978. Birds and army ants. Annual Review of Ecology and Systematics 9:243-263. van der Zande, A. N., W. J. ter Keurs, and W J. van der Weijden. 1980. The impact of roads on the densities of four bird species in an open field habitat: evidence of a long-distance effect. Biological Conservation 18:299-321. Appendix 1. Number of within-forest and road-crossing movements recorded for each species and guild of Amazonian understory bird. Appendix 1. (continued) species/guild Within-forest movements Army-ant followers Pithys albifrons 103 Gymnopithys rufigula 36 Dendrocincla merula 8 subtotal 147 Solitary understory species Hylophylax poedlinota 57 Hylophylax naevia 1 Microbates collaris 7 Microcerculus bamhla 1 Automolus rubiginosus 4 Dendrocincla fuliginosa 2 Platyrinchus saturatus 9 Myrmotherula guttata 2 Frederickena viridis 1 Onychorhynchus coronatus 0 Malacoptila fusca 4 subtotal 88 Mixed-species flocks Thamnomanes ardesciacus 23 Thamnomanes caesius 15 Myrmotherula axillaris 6 Myrmotherula gutturalis 15 Myrmotherula longipennis 5 Myrmotherula menetriesii 1 Hylophilus ochraceiceps 4 Myiobius barbatus 6 Automolus infuscatus 10 Philydor erythrocercus 1 Xenops minutus 4 Deconychura stictolaema 12 Deconychura longicauda 1 Glyphorynchus spirurus 46 Xiphorhynchus pardalotus 16 subtotal 165 Road-crossing inovements 20 6 3 29 5 0 0 1 0 1 0 0 0 0 0 7 .3 2 1 2 0 0 2 2 0 0 1 1 21 12 51 Within-forest Road-crossing Species/guild movements movements Terrestrial species Arremon tacitumis 1 0 Corythopis torquata 0 1 Formicarius colma 9 1 Formicarius analis 0 0 Myrm eciza ferruginea 5 0 Sclerurus caudacutus 2 0 Scierurus ru?guiaris 1 0 subtotal 18 2 Edge/gap species GaJbuJu albirostris 11 5 Trogon rufus 1 0 Hypocnemis cantator 5 2 Percnostola rufifrons 15 15 Automolus ochrolaemus 1 3 Cyanocompsa cyanoides 3 0 Tachyphonus surinamus 2 1 Thryothorus coraya 2 0 subtotal 41 26 Frugivore species Geotrygon montana 3 2 Piprapipra 36 16 Pipra serena 2 3 Schiffomis turdinus 7 0 Turdus albicollis 19 1 subtotal 67 22 Rare guilds not included in guild-level analyses solitary mid-story species Momotus momota 3 0 Buceo capensis 1 0 Helexetastes perrotii 0 1 Omnivore species Mionectes macconnelli 8 3 Canopy species Tachyphonus cristatus 0 1 Nocturnal carnivore species Otus watsoni 1 0 All birds 538 142 Conservation Biology Volume 18, No. 4, August 2004