Recruitment curves for eight large introduced tree species planted in the Muliwai region of Hawai‘i Island during the 1930s. Five species are naturalized and regenerating with varying success, and three species have failed to regenerate (although these three species are naturalized in other areas of Hawai‘i [ Wagner et al. 1999]). On the x -axis, the highest median dbh (diameter at breast height) category includes all stems > 50 cm. Note the different scales on the y -axes. 

Contexts in source publication

Context 1

... shade-tolerant species, the failure of Cibotium spp. to succeed in the planted forest suggests a different mechanism than the niche collapse observed for M. polymorpha . A probable cause is the high abundance of feral ungulates, which feed on the caudices of tree ferns. In areas where tree ferns are quite common, such as the relict forest studied here, ungulates typically do not cause enough damage to lead to declining Cibotium spp. abundance but may play a role in limiting their recruitment (Sweetapple and Nugent 2004). With the absence of dominant species such as M. polymorpha and Cibotium spp., Hawaiian rain forests of the future will differ dramatically from their current composition. But will they lose all native tree species? I found evidence that two common native understory species, P. hawaiiensis and P. odorata, may continue to persist. Each of these species reached its maximum size beneath the plantation (according to Wagner et al. 1999) and appeared to be regenerating suc- cessfully based on recruitment curves of all individuals reaching 1.3 m in height (Figure 2) and detectable levels of seedlings (Table 3). Notably, I found more stems/ha for both species in the planted compared with the relict forest, and P. odorata was significantly more abundant in the planted forest. In a pre- vious study, Mascaro et al. (2008) found that P. odorata was more abundant in novel (nonplanted) forests dominated by F. moluccana and Casuarina equisetifolia (ironwood) than it was in nearby relict native forests in lower Puna on Hawai‘i Island. Those authors also found that P. hawaiiensis was the most abundant native understory component in a survey of 46 novel forests dominated by introduced species across Hawai‘i Island. Combined with the results presented here, these data suggest that both P. odorata and P. hawaiiensis are established native components of introduced forests on Hawai‘i. The success of P. odorata and P. hawaiiensis compared with M. polymorpha is likely due in part to their high tolerance of shade. The shift from dominance by M. polymorpha to introduced trees, particularly the gymnosperm A. columnaris, which is reproducing prolifi- cally in the plantation, typically results in a much darker understory and a light regime intolerable to M. polymorpha (Hughes and Denslow 2005, Ostertag et al. 2008). Although I did not measure light availability, total basal area in the planted forest (71 e 8 m 2 /ha) was 65% higher than in the relict forest (43 e 4 m 2 /ha) (data not shown), suggesting substantially higher leaf area and thus light interception. Shade tolerance did not guarantee success, however. In contrast to the success in the plantation of P. odorata and P. hawaiiensis, two other functionally similar species ( A. platyphyllum and M. lessertiana ) have failed to succeed in the plantation. Several non- mutually exclusive mechanisms could be re- sponsible for the relative success of some native understory tree species over others in the planted forest, including herbivore pressure (Cabin et al. 2000, Sweetapple and Nugent 2004, Joe and Daehler 2008), the availability of pollinator and dispersal agents (Foster and Robinson 2007), and seed predation (Wagner et al. 1999). Each of these mechanisms has been documented to play a role in changing community compositions on Hawai‘i. Throughout the Hawaiian Islands, for instance, herbivory by ungulates and slugs di- rectly impedes the regeneration of certain native species, with the impacts distributed inequitably among species (Cabin et al. 2000, Sweetapple and Nugent 2004, Joe and Daehler 2008). On Maui, some native tree species are able to utilize introduced frugivorous birds for dispersal in a planted forest, favoring their recruitment over that of other native tree species (Foster and Robinson 2007). In this case the native frugivores that formerly dispersed the native tree species are extinct. Finally, seed predators such as introduced rats preferentially target certain native tree species but not others; they are an extreme threat to the endangered Pritchardia spp. (Wagner et al. 1999). Additional research is needed to determine the specific mechanisms behind the relative success of native understory tree species in new forests on Hawai‘i. Although this study targeted one of the oldest planted forests on Hawai‘i, succession is continuing, and the prospects for native species could change. As the initial cohort of planted trees begins to die, gaps and landslide disturbances will create new regeneration opportunities that will likely cause the overstory composition to change. At a minimum, this will change the relative abundances of the introduced species that dominate the forest. For example, three of the five most dominant introduced species, M. quinquenervia, E. robusta, and G. robusta, have produced virtually no recruits in 80 yr (Table 1, Figure 3 F–H ), while the remaining two, A. columnaris and C. glauca, are regenerating strongly (Table 1, Figure 3 A–B ). In addition, several introduced pioneer species that were not part of the initial plantation and currently occur in low abundance in the plantation, such as Cecropia obtusifolia (trumpet tree), Melochia umbellata (melochia), and Schefflera actinophylla (octopus tree), may take advantage of these opportunities as they do in canopy gaps elsewhere on the island (Zimmerman et al. 2008). Although it is present at low abundance in the planted forest, the native dominant M. polymorpha is unlikely to take advantage of these regeneration opportunities due to particularly low abundance of low-statured stems and seedlings that would be likely to benefit from gaps. Furthermore, M. polymorpha has failed to succeed in a myriad of light and nutrient conditions in both lowland native and exotic- dominated forests across the island (Hughes and Denslow 2005, Mascaro et al. 2008, Zimmerman et al. 2008). As suggested by Mueller-Dombois (2008), monodominance in many ecosystems by M. polymorpha is in decline, and a new natural dynamic has taken course, driven primarily by introduced species with a few natives remaining. Left unaltered, the planted area will gradu- ally transition from an identifiable plantation to something more resembling a novel ecosystem (sensu Hobbs et al. 2006), one in which introduced species dominate but where human intervention is absent. In this context, the Muliwai region may have considerable value as a predictor of future forests; it is much older than other novel ecosystems that are rapidly increasing on Hawai‘i Island (Mascaro et al. 2008). On the other hand, if enhanced abundance of native species is de- sirable, the future for natives could also be altered deliberately by land management. A traditional effort to remove all introduced species would be enormously expensive and likely counterproductive. Removing nearly the entire plantation would create a very high light environment that, on other parts of Hawai‘i Island, has resulted in complete dominance by introduced pioneer species (Mascaro et al. 2008). An alternative that has proved to be cost-effective elsewhere may be to target control efforts in the understory of the plantation while keeping the canopy trees intact (see Kueffer et al. 2007). In this way, forest succession could be steered gently toward higher native abundance, without causing the type of interruption that would surely benefit opportunistic introduced species. A limitation of this study is that the relict native forest is far from pristine, and thus although it serves as a baseline for the abundance of native tree species that might be found in a typical lowland native forest today, all of which contain introduced species (Zimmerman et al. 2008), it cannot act as a baseline for the richness and abundance of native species in the Muliwai area before human contact. However, because this precontact baseline is fundamentally unknowable, the relict forest provides the best available refer- ence community for the purpose of under- standing contemporary community shifts. Although native species (notably M. polymorpha ) are unlikely to attain dominance in the planted forest studied here, the shade- tolerant species P. odorata and P. hawaiiensis are currently found at comparable or (in the case of P. odorata ) higher abundance in the planted forest compared with an adjacent relict native forest. The continued presence and strong recruitment of these species after 80 yr of succession with a wide diversity of introduced species, both planted and nonplanted, suggest that they are likely to persist in this region of Hawai‘i Island for some time. I thank S. Bergfeld and the Hawai‘i Division of Forestry and Wildlife for granting site access, J. Brown and R. McDowell for field assistance, and two anonymous reviewers for comments on an early draft of the manu- script. Asner, G. P., R. F. Hughes, P. M. Vitousek, D. E. Knapp, T. Kennedy-Bowdoin, J. Boardman, R. E. Martin, M. Eastwood, and R. O. Green. 2008. Invasive plants transform the three-dimensional structure of rain forests. Proc. Natl. Acad. Sci. U.S.A. 105:4519–4523. Cabin, R. J., S. G. Weller, D. H. Lorence, T. W. Flynn, A. K. Sakai, D. Sandquist, and L. J. Hadway. 2000. Effects of long- term ungulate exclusion and recent alien species control on the preservation and restoration of a Hawaiian tropical dry forest. Conserv. Biol. 14:439–453. Constantinides, M., and R. J. Cannarella. 1999. A timber inventory of the Waiakea Timber Management Area. The State of Hawai‘i, Division of Forestry and Wildlife, Honolulu. Denslow, J. S. 1980. Gap partitioning among tropical rainforest trees. Biotropica 12:47– 55. Egler, F. E. 1942. Indigene versus alien in the development of arid Hawaiian vegetation. Ecology 23:14–23. Ewel, J. J., and F. E. Putz. 2004. A place for alien species in ecosystem restoration. Front. Ecol. Environ. 2:354–360. Foster, J. T., and S. K. Robinson. 2007. Introduced birds and the fate of Hawaiian ...

Context 2

... stems within 12.62 m radius and a bearing of 0 to 90 ), only individuals b 10 cm dbh were measured thereafter. All species abundance data were then weighted based on the area sampled. This method was chosen to preserve the integrity of species richness and basal area measurements, while still providing a useful picture of the full range of size classes for all species. No native species reached this threshold of abundance. Within each plot, I also recorded the identity and abundance of seedlings < 1.3 m in height in a 1 m by 5 m subplot between 5 and 10 m north of the plot center and extending 1 m east of a bearing of 0 . The placement was intended to avoid trampling at the plot center. Due to a high number of zero values, species abundances were not normally distributed. I compared species abundances between planted and relict forest types using non- parametric Mann-Whitney U tests (www .r-project.org). I created histograms of species abundances by binning stems into 20 2.5 cm dbh size classes from 0 to 50 cm dbh for each species in a given forest type. I graphed the results by centering each data column around the median dbh of each size class (e.g., the first size class ranged from 0 to 2.49 cm dbh, inclusive, and was centered at 1.25 cm dbh when graphed). Each graph also included a composite bin category for all stems > 50 cm dbh, which was centered at 51.25 cm when graphed. After 80 yr of succession in a noncommercial plantation, native species constituted just 4.5% of basal area and 12.1% of stem density (Tables 1 and 2). Of nine native tree species found in all plots, two species were rare in both forest types: Diospyros sandwicensis ( lama ) was found in only one plot in each forest, and Wikstroemia spp. (‘ akia ) were found in only one relict forest plot. Five native species that were common in relict native forest were either absent or rare in planted forest ( Antidesma platyphyllum [ hame ], Cibotium glaucum [ hapu‘u ], Cibotium menziesii [ hapu‘u ], M. polymorpha , and Myrsine lessertiana [ k ̄ lea ]) (Figure 1). However, two native tree species ( Psychotria hawaiiensis [ k ̄ piko ] and Psydrax odorata [ alahe‘e ]) were common in the planted forest (Table 2). The abundance of P. hawaiiensis did not differ significantly between the relict and planted forests ( U 1⁄4 248, P 1⁄4 : 7982, n 1 1⁄4 26, n 2 1⁄4 20), and P. odorata was significantly more abundant in the planted forest ( U 1⁄4 146 : 5, P 1⁄4 : 0074, n 1 1⁄4 26, n 2 1⁄4 20) (Figure 1). These low-stature tree species each exceeded 15 cm dbh and had strong recruitment curves in both relict and planted forest, suggesting that they are completing their life cycle (Figure 2). By contrast, the recruitment curves of the native dominant M. polymorpha suggest that only scattered individuals have succeeded in the planted forest and indicate relatively poor recruitment of young individuals in the relict forest. Both P. odorata and P. hawaiiensis were also found at the seedling stage in the planted forest, although at relatively low abundances compared with the introduced species (Table 3). These were the only native species found at the seedling stage in the planted forest. In addition to the lack of native species in the planted forest, the relict forest was domi- nated by roughly equal parts native (55% of basal area) and introduced species (45%) (Table 1). The native component was nearly all M. polymorpha (50% of basal area), and the introduced component consisted primarily of the short-stature introduced trees Psidium guajava (common guava) and Cordyline fruticosa ( ti ). Notably, both of these introduced species were less abundant in the planted forest both as stems b 1.3 m in height (Table 2) and as seedlings (Table 3). Similarly, the introduced C. hirta attained the highest relative abundance in the relict forest and was less abundant in the plantation (Table 2). The plantation was dominated by very large tree species widely planted in the 1930s throughout Hawai‘i (Woodcock 2003) (Table 1). Areas of level grade, to which most of my sampling was constrained, were planted primarily with a mix of Falcataria moluccana (al- bizia), Melaleuca quinquenervia (paperbark), Eucalyptus robusta (swamp mahogany), Fraxi- nus uhdei (tropical ash), and Araucaria columnaris (Cook pine). Ridges and slopes were primarily planted with Casuarina glauca (ironwood) and Grevillea robusta (silk oak), and steep valleys and watercourses were primarily planted with Terminalia myriocarpa ( jhalna). The plantation also included a minor component of large ornamental tree species (e.g., Ficus elastica [rubber fig], Taxodium distichum [bald cypress]), and patches of fruit trees (e.g., Macadamia integrifolia [macadamia nut], Citrus limon [lime], and Coffea arabica [coffee]) (Appendix). Recruitment curves suggest that some planted species are recruiting strongly (e.g., A. columnaris and C. glauca ), but others are unable to recruit (e.g., E. robusta and G. robusta ) (Figure 3). Lowland Hawaiian forests have entered a period of rapid change, and it is an open question whether native tree species will be maintained. After 80 yr of succession, I found that the dominant native canopy tree ( M. polymorpha ), as well as the dominant native midcanopy tree ferns ( Cibotium spp.), have thus far been unable to succeed in the plantation. Mueller-Dombois (2008) made the case that the native Hawaiian tree flora lacks dominant canopy species with the clear succes- sional roles we typically ascribe to tropical tree species, such as r-selected, fast-growing pioneer species, and K-selected, slow- growing shade-tolerant species (Denslow 1980). For example, M. polymorpha tends toward high intolerance of shade and slow growth—traits that are likely causing its niche space to collapse in the presence of hundreds of new competitors. Observational studies have found that M. polymorpha dominance in lowland (and some midelevation) regions of Hawai‘i is eroding rapidly and lacking in regeneration (Hughes and Denslow 2005, Mascaro et al. 2008, Ostertag et al. 2008, Zimmerman et al. 2008), and exper- imental studies have found that M. polymorpha cannot compete for resources against introduced tree species in lowland wet conditions (Funk and Vitousek 2007, Ostertag et al. 2009). In support of these conclusions, I found that M. polymorpha was largely unable to recruit into a planted forest after 80 yr of succession despite an ample seed source from a nearby relict native forest, as evidenced by a very low abundance of low-statured stems and lack of seedlings. Furthermore, I found that M. polymorpha had poor recruitment in the relict forest, suggesting that continued succession there will lead to a gradual loss of this dominant canopy tree. Metrosideros poly- morpha often depends on gaps to maintain dominance in closed-canopy forest (Mueller- Dombois 2008) and might reasonably be expected to colonize gaps in the plantation. However, because low-statured stems and seedlings of M. polymorpha were rare and absent, respectively, I suggest that there is no cohort of available M. polymorpha individuals available to take advantage of such opportunities. Metrosideros polymorpha may be able to recruit in forests dominated by introduced species in limited situations with some combi- nation of dry conditions, high light, low nutrient availability, or high elevation (i.e., with lower pressure from introduced species). Mueller-Dombois (2008) observed strong recruitment of M. polymorpha along with several introduced tree species near Kalapana (Hawai‘i Island). Although the area is at low elevation, lavas are young and strongly nutrient limited, providing a high-stress and high- light environment that may favor M. polymorpha . Harrington and Ewel (1997) and Constantinides and Cannarella (1999) found that M. polymorpha recruited beneath P 25-yr-old plantings of Eucalyptus spp., Toona ciliata (red cedar), and F. uhdei in the Wai ̄kea Timber Management Area on Hawai‘i Island. However, recruitment occurred primarily above 350 m elevation in an area of planted forest surrounded entirely by undisturbed native forest. The Wai ̄kea Timber Management Area is located on strongly nutrient-limited lava that was never in a nonforest land-use state (contrary to most plantings on Hawai‘i Island), suggesting that the Wai ̄kea Timber Management Area may not be representative of M. polymorpha regeneration under introduced species even at higher elevation. In addition, current work near the Wai ̄kea Timber Management Area suggests that M. polymorpha recruitment is rapidly declining as Psidium cattleianum (strawberry guava) ad- vances into higher-elevation forests ( J. Denslow, U.S. Forest Service, pers. comm.). On O‘ahu, Woodcock et al. (1999) found that M. polymorpha was a common understory component of a Melaleuca forest planted in the 1930s (275 m elevation). Woodcock et al. speculated that the high light availability in the ridge-top conditions may have allowed M. polymorpha to recruit into the plantation. Even after 70 yr of growth, however, no M. polymorpha trees were taller than 1.5 m. Two common tree fern species ( Cibotium spp.) that characterize nearly all native rain forests on Hawai‘i were nearly absent in the planted forest. These species were also nearly absent in novel (nonplanted) lowland forests surveyed by Mascaro et al. (2008), but like M. polymorpha, they recruited strongly in high-elevation areas of the Wai ̄kea Timber Management Area (Harrington and Ewel 1997, Constantinides and Cannarella 1999). As late-successional shade-tolerant species, the failure of Cibotium spp. to succeed in the planted forest suggests a different mechanism than the niche collapse observed for M. polymorpha . A probable cause is the high abundance of feral ungulates, which feed on the caudices of tree ferns. In areas where tree ferns are quite common, such as the relict forest studied here, ungulates typically do not cause ...

Context 3

... ̄kea Timber Management Area (Harrington and Ewel 1997, Constantinides and Cannarella 1999). As late-successional shade-tolerant species, the failure of Cibotium spp. to succeed in the planted forest suggests a different mechanism than the niche collapse observed for M. polymorpha . A probable cause is the high abundance of feral ungulates, which feed on the caudices of tree ferns. In areas where tree ferns are quite common, such as the relict forest studied here, ungulates typically do not cause enough damage to lead to declining Cibotium spp. abundance but may play a role in limiting their recruitment (Sweetapple and Nugent 2004). With the absence of dominant species such as M. polymorpha and Cibotium spp., Hawaiian rain forests of the future will differ dramatically from their current composition. But will they lose all native tree species? I found evidence that two common native understory species, P. hawaiiensis and P. odorata, may continue to persist. Each of these species reached its maximum size beneath the plantation (according to Wagner et al. 1999) and appeared to be regenerating suc- cessfully based on recruitment curves of all individuals reaching 1.3 m in height (Figure 2) and detectable levels of seedlings (Table 3). Notably, I found more stems/ha for both species in the planted compared with the relict forest, and P. odorata was significantly more abundant in the planted forest. In a pre- vious study, Mascaro et al. (2008) found that P. odorata was more abundant in novel (nonplanted) forests dominated by F. moluccana and Casuarina equisetifolia (ironwood) than it was in nearby relict native forests in lower Puna on Hawai‘i Island. Those authors also found that P. hawaiiensis was the most abundant native understory component in a survey of 46 novel forests dominated by introduced species across Hawai‘i Island. Combined with the results presented here, these data suggest that both P. odorata and P. hawaiiensis are established native components of introduced forests on Hawai‘i. The success of P. odorata and P. hawaiiensis compared with M. polymorpha is likely due in part to their high tolerance of shade. The shift from dominance by M. polymorpha to introduced trees, particularly the gymnosperm A. columnaris, which is reproducing prolifi- cally in the plantation, typically results in a much darker understory and a light regime intolerable to M. polymorpha (Hughes and Denslow 2005, Ostertag et al. 2008). Although I did not measure light availability, total basal area in the planted forest (71 e 8 m 2 /ha) was 65% higher than in the relict forest (43 e 4 m 2 /ha) (data not shown), suggesting substantially higher leaf area and thus light interception. Shade tolerance did not guarantee success, however. In contrast to the success in the plantation of P. odorata and P. hawaiiensis, two other functionally similar species ( A. platyphyllum and M. lessertiana ) have failed to succeed in the plantation. Several non- mutually exclusive mechanisms could be re- sponsible for the relative success of some native understory tree species over others in the planted forest, including herbivore pressure (Cabin et al. 2000, Sweetapple and Nugent 2004, Joe and Daehler 2008), the availability of pollinator and dispersal agents (Foster and Robinson 2007), and seed predation (Wagner et al. 1999). Each of these mechanisms has been documented to play a role in changing community compositions on Hawai‘i. Throughout the Hawaiian Islands, for instance, herbivory by ungulates and slugs di- rectly impedes the regeneration of certain native species, with the impacts distributed inequitably among species (Cabin et al. 2000, Sweetapple and Nugent 2004, Joe and Daehler 2008). On Maui, some native tree species are able to utilize introduced frugivorous birds for dispersal in a planted forest, favoring their recruitment over that of other native tree species (Foster and Robinson 2007). In this case the native frugivores that formerly dispersed the native tree species are extinct. Finally, seed predators such as introduced rats preferentially target certain native tree species but not others; they are an extreme threat to the endangered Pritchardia spp. (Wagner et al. 1999). Additional research is needed to determine the specific mechanisms behind the relative success of native understory tree species in new forests on Hawai‘i. Although this study targeted one of the oldest planted forests on Hawai‘i, succession is continuing, and the prospects for native species could change. As the initial cohort of planted trees begins to die, gaps and landslide disturbances will create new regeneration opportunities that will likely cause the overstory composition to change. At a minimum, this will change the relative abundances of the introduced species that dominate the forest. For example, three of the five most dominant introduced species, M. quinquenervia, E. robusta, and G. robusta, have produced virtually no recruits in 80 yr (Table 1, Figure 3 F–H ), while the remaining two, A. columnaris and C. glauca, are regenerating strongly (Table 1, Figure 3 A–B ). In addition, several introduced pioneer species that were not part of the initial plantation and currently occur in low abundance in the plantation, such as Cecropia obtusifolia (trumpet tree), Melochia umbellata (melochia), and Schefflera actinophylla (octopus tree), may take advantage of these opportunities as they do in canopy gaps elsewhere on the island (Zimmerman et al. 2008). Although it is present at low abundance in the planted forest, the native dominant M. polymorpha is unlikely to take advantage of these regeneration opportunities due to particularly low abundance of low-statured stems and seedlings that would be likely to benefit from gaps. Furthermore, M. polymorpha has failed to succeed in a myriad of light and nutrient conditions in both lowland native and exotic- dominated forests across the island (Hughes and Denslow 2005, Mascaro et al. 2008, Zimmerman et al. 2008). As suggested by Mueller-Dombois (2008), monodominance in many ecosystems by M. polymorpha is in decline, and a new natural dynamic has taken course, driven primarily by introduced species with a few natives remaining. Left unaltered, the planted area will gradu- ally transition from an identifiable plantation to something more resembling a novel ecosystem (sensu Hobbs et al. 2006), one in which introduced species dominate but where human intervention is absent. In this context, the Muliwai region may have considerable value as a predictor of future forests; it is much older than other novel ecosystems that are rapidly increasing on Hawai‘i Island (Mascaro et al. 2008). On the other hand, if enhanced abundance of native species is de- sirable, the future for natives could also be altered deliberately by land management. A traditional effort to remove all introduced species would be enormously expensive and likely counterproductive. Removing nearly the entire plantation would create a very high light environment that, on other parts of Hawai‘i Island, has resulted in complete dominance by introduced pioneer species (Mascaro et al. 2008). An alternative that has proved to be cost-effective elsewhere may be to target control efforts in the understory of the plantation while keeping the canopy trees intact (see Kueffer et al. 2007). In this way, forest succession could be steered gently toward higher native abundance, without causing the type of interruption that would surely benefit opportunistic introduced species. A limitation of this study is that the relict native forest is far from pristine, and thus although it serves as a baseline for the abundance of native tree species that might be found in a typical lowland native forest today, all of which contain introduced species (Zimmerman et al. 2008), it cannot act as a baseline for the richness and abundance of native species in the Muliwai area before human contact. However, because this precontact baseline is fundamentally unknowable, the relict forest provides the best available refer- ence community for the purpose of under- standing contemporary community shifts. Although native species (notably M. polymorpha ) are unlikely to attain dominance in the planted forest studied here, the shade- tolerant species P. odorata and P. hawaiiensis are currently found at comparable or (in the case of P. odorata ) higher abundance in the planted forest compared with an adjacent relict native forest. The continued presence and strong recruitment of these species after 80 yr of succession with a wide diversity of introduced species, both planted and nonplanted, suggest that they are likely to persist in this region of Hawai‘i Island for some time. I thank S. Bergfeld and the Hawai‘i Division of Forestry and Wildlife for granting site access, J. Brown and R. McDowell for field assistance, and two anonymous reviewers for comments on an early draft of the manu- script. Asner, G. P., R. F. Hughes, P. M. Vitousek, D. E. Knapp, T. Kennedy-Bowdoin, J. Boardman, R. E. Martin, M. Eastwood, and R. O. Green. 2008. Invasive plants transform the three-dimensional structure of rain forests. Proc. Natl. Acad. Sci. U.S.A. 105:4519–4523. Cabin, R. J., S. G. Weller, D. H. Lorence, T. W. Flynn, A. K. Sakai, D. Sandquist, and L. J. Hadway. 2000. Effects of long- term ungulate exclusion and recent alien species control on the preservation and restoration of a Hawaiian tropical dry forest. Conserv. Biol. 14:439–453. Constantinides, M., and R. J. Cannarella. 1999. A timber inventory of the Waiakea Timber Management Area. The State of Hawai‘i, Division of Forestry and Wildlife, Honolulu. Denslow, J. S. 1980. Gap partitioning among tropical rainforest trees. Biotropica 12:47– 55. Egler, F. E. 1942. Indigene versus alien in the development of arid Hawaiian vegetation. Ecology 23:14–23. Ewel, J. J., and F. E. Putz. 2004. A place for alien species in ecosystem restoration. Front. Ecol. ...