This annotated bibliography is provided in order to assess the achievements and gaps in the literature on vegetative identification of woody plants in the tropics. The bibliography includes 258 references divided into general references (48), pantropical and multi-continental keys (4), Africa (46), Asia (53), Australia and New Guinea (21), and the Neotropics (86). Vegetative keys are not easy shortcuts to the identification of woody tropical plants to species. Identification of sterile material requires as much, if not more, training as that required for using keys that emphasize floral and fruit characters. Nevertheless, the literature surveyed in this bibliography may be helpful to biologists seeking references for the geographic region in which they work, models for the construction of vegetative keys and field guides, and/or the types of characters that should be observed in the field.

For many soils of the tropics, inputs of organic materials are essential to sustain soil fertility and crop production. Research in the quality of organic inputs, a key factor controlling rates of decomposition and nutrient release, continues to guide selection and use of organic materials as nutrient sources. The relationship between decomposition patterns and the quality parameters of the fresh leaves of six agroforestry species: Sesbania sesban, Croton megalocarpus, Calliandra calothyrsus, Tithonia diversifolia, Lantana camara, and Senna spectabilis, was investigated in a litterbag study over a period of 77 days in the highlands of western Kenya. The litterbags were buried 1 cm below the soil surface and covered with soil of ca 1 cm thickness. Percent leaf mass and total N and P that remained with time strongly correlated with total P and C/P ratio (R² = 0.60–0.90) during the first 35 days of study; but afterwards, correlation was stronger with the initial soluble polyphenolics (Pp)/P ratio (R² = 0.69–0.92) than with total P and C/P ratio. Loss of leaf mass and release of N and P followed the exponential function, y_t = y₀ * e^−kt, from which the specific decay rate constants (k) were calculated for loss of leaf mass (k_B) and release of N (k_N) and P (k_P). Among the plant species, the k values were lowest in Calliandra with k_B = 0.012/d, k_N = 0.017/d and k_P = 0.044/d. Lantana had the highest k values with k_B = 0.067/d and k_P = 0.119/d, but the highest k_N value of 0.109/d occurred in Tithonia. The k_B values for all organic materials were lower than their corresponding k_N and k_P values, suggesting that leaching of N and P from litters may have augmented the microbial mineralization of N and P. There was a strong correlation between the k_B, k_N, and k_P values and total P (r = 0.82–0.96; P < 0.01), but not total N, lignin (LIG), or Pp. Rates of N and P release followed the general trend: Tithonia > Senna > Lantana > Sesbania > Croton > Calliandra. The results indicated that, among the quality parameters studied, total P is the most important factor controlling rate of decomposition and N and P release from organic inputs in the area of study.

In temperate forested streams, fruit from riparian trees is generally a minor and seasonal component of the allochthonous detritus. In contrast, riparian fruit input to tropical streams is often high and continuous. Detrital fruit is abundant in some forested Hawaiian streams compared to other forms of riparian detritus, and rates of leaf litter processing by macroscopic invertebrates are very low. These observations suggested that fruit is an important food resource for detritivores. A microcosm system was used to measure the rates at which two common detrital fruits, guava and mango, were processed by two common detritivores, the prawn Macrobrachium lar and the gastropod Tarebia granifera. Comparisons of fruit weight loss rates normalized by detritivore weight indicated that M. lar processed guavas at significantly higher rates than T. granifera; differences in rates of mango processing by M. lar and T. granifera were not significant. Microcosms containing both M. lar and T. granifera were used to test for interactions between the invertebrates that affected rates of mango processing. No interspecific interactions were detected. A field study was conducted in Kaiwiki Stream, Island of Hawaii, to determine rates of detrital fruit input and export. Detrital fruit was supplied to the study area year-round, with peaks corresponding to summer and autumn fruiting seasons. Guavas and mangos accounted for 85 percent of the fruit biomass entering the stream and 92 percent of the fruit exported from the stream. Mean daily export rates of guava were 7 percent of input, and export rates of mango were 5 percent of input. These measurements suggested that most of the fruit entering the stream is retained and comprises a substantial food resource for detritivores. Comparisons of the biomass-specific rates at which M. lar and T. granifera processed mangos and guavas with the rates at which mangos and guavas entered Kaiwiki Stream suggested that these invertebrates can process most of the detrital fruit in the stream.

In 1990 and 1991, Samoa was struck by two cyclones, Ofa and Val. In the Tafua Rain Forest Preserve on the island of Savai'i, one part of the forest also burned after the first cyclone. Here we report on patterns of regeneration and changes in tree species composition in the Tafua lowland rain forest after five years of recovery from cyclone and fire disturbance. In the unburned area, tree canopy cover increased from 27 percent after the last cyclone to 58 percent, and in the burned area from below 12 to 49 percent. Nine of the ten most common tree species decreased in relative abundance in the entire forest after the last cyclone. One fast growing pioneer species, Macaranga harveyana, now makes up 42 percent of the total number of trees (>5 cm DBH) in the unburned area and 86 percent in the burned area. Large interspecific differences occur in size distribution and there are at least four distinguishable regeneration patterns, which may be related to shade tolerance. Mean number of species per plot was generally higher in the unburned area than in the burned area, while the Shannon evenness index was higher in the unburned than in the burned area only for trees above 1 cm DBH. Species with fruits known to be fed upon by birds and/or bats generally made up a larger proportion of all trees in the burned than in the unburned area. In contrast to other studies of post-cyclone regeneration, in which recovery is often rapid due to resprouting of trees, recovery in the Tafua forest was a slow process with regeneration more dependent on vertebrate seed dispersal than on resprouting.

Although lack of seeds can limit forest recovery in abandoned agricultural lands, few studies have documented the patterns of seed rain and soil seed bank in active pastures. To determine if lack of seeds is a limiting factor, we studied the woody species composition of the seed rain and soil seed bank in an active pasture in Barranquitas, Puerto Rico. In addition, fruit production of 43 common shrub and tree species was monitored in the surrounding secondary forest. Seed rain was monitored for one year at −2, 0, 2, 4, 8, 16, and 32 m from the forest edge along six transects. The soil seed bank was sampled in January/February 1995 and October/November 1995 at the same distances from the forest edge along ten transects. Of the 35 species that produced fruits in the forest, 14 species were detected in the seed rain study and only 0.3 percent of the seeds and 3 species dispersed to more than 4 m from the forest edge. Two of the three species were dispersed by wind. The seed bank study showed a similar pattern with a dramatic decrease in seedling density and species richness with distance from the forest edge. More than 50 percent of the seedlings in the seed bank study were Trema lamarckianum. This study demonstrated that few seeds disperse into the pastures and even when a rare dispersal event occurs, species do not accumulate because of short-term seed viability and, possibly, high seed predation. Studies of early secondary forest regeneration have shown that in some cases, forest can recover rapidly in abandoned agricultural lands; but our results suggest that only a small subset of the forest species will contribute to the initial recovery process.

In hardwood subtropical forests of southern Florida, nonnative vines have been hypothesized to be detrimental, as many species form dense “vine blankets” that shroud the forest. To investigate the effects of nonnative vines in post-hurricane regeneration, we set up four large (two pairs of 30 × 60 m) study areas in each of three study sites. One of each pair was unmanaged and the other was managed by removal of nonnative plants, predominantly vines. Within these areas, we sampled vegetation in 5 × 5 m plots for stems 2 cm DBH (diameter at breast height) or greater and in 2 × 0.5 m plots for stems of all sizes. For five years, at annual censuses, we tagged and measured stems of vines, trees, shrubs and herbs in these plots. For each 5 × 5 m plot, we estimated percent coverage by individual vine species, using native and nonnative vines as classes. We investigated the hypotheses that: (1) plot coverage, occurrence and recruitment of nonnative vines were greater than that of native vines in unmanaged plots; (2) the management program was effective at reducing cover by nonnative vines; and (3) reduction of cover by nonnative vines improved recruitment of seedlings and saplings of native trees, shrubs, and herbs. In unmanaged plots, nonnative vines recruited more seedlings and had a significantly higher plot-cover index, but not a higher frequency of occurrence. Management significantly reduced cover by nonnative vines and had a significant overall positive effect on recruitment of seedlings and saplings of native trees, shrubs and herbs. Management also affected the seedling community (which included vines, trees, shrubs, and herbs) in some unanticipated ways, favoring early successional species for a longer period of time. The vine species with the greatest potential to “strangle” gaps were those that rapidly formed dense cover, had shade tolerant seedling recruitment, and were animal-dispersed. This suite of traits was more common in the nonnative vines than in the native vines. Our results suggest that some vines may alter the spatiotemporal pattern of recruitment sites in a forest ecosystem following a natural disturbance by creating many very shady spots very quickly.

Postfire vegetation development was studied at two recent mountaintop burns in the Serra do Caparaó, Espírito Santo/Minas Gerais, and the Serra dos Órgãos, Rio de Janeiro, Brazil. Both fires occurred within the páramo-like “campos de altitude”, a system of orographic grass- and scrublands restricted to the highest peaks of southeast Brazil. Data collected included pre- and postfire heights and densities of shrubs and trees, and cover of all taxa. Slopes of different aspect and altitude were sampled at each site to evaluate the effects of varying physical environment on regeneration. Rapid vegetative regeneration was common among shrubs and bamboo, with most taxa surviving fire. Regeneration and postfire colonization rates varied among species, and appeared to depend both on physical variables associated with different slopes, aspects and altitudes, and biotic variables. Most plant species in the campos de altitude show some form of evolutionary adaptation to fire. The distribution of species and plant populations across the landscapes of the campos de altitude appears to be largely the result of fire and its interactions with the biota, local topography, and climate. Results suggest that fire is an integral part of the ecology in these humid, seasonally dry ecosystems. Successful management of these unique and highly threatened systems requires a detailed understanding of the fire regime and its role in structuring biotic communities.

Results of a shadehouse experiment and a field study with seedlings and saplings of Inga leiocalycina Willd. (Leguminoseae), a canopy tree of American lowland tropical wet forests, indicated that plant light environment affects both (1) the degree of root colonization by vesicular-arbuscular mycorrhizal (VAM) fungi and (2) the development of VAM structures in the root. Among shadehouse-grown seedlings, VAM colonization increased with increasing light intensity, from a mean of 20 percent (21.5 cm colonized root) at a light level approximating forest understory environments to 47 percent (233.1 cm) at a light level found in centers of treefall gap environments. In contrast, VAM colonization levels were similar between gap and understory habitats among field-collected sapling roots; however, the anatomy of the symbiosis differed between light environments. Roots of light gap saplings had significantly more intracellular fungal coils than understory sapling roots. At subsaturating levels for photosynthesis, light availability appeared to be a limiting resource for this symbiosis, restricting the extent of VAM fungal colonization and development of symbiotic structures. Despite this light-mediated limitation, woody seedlings showed evidence of VAM fungi in 20 to 30 percent of their roots at light levels typical of the tropical wet forest ecosystem in general. I consider these results in the context of forest dynamics, and I suggest that, for woody seedlings, maintaining moderate levels of VAM in understory environments is of long-term benefit because it enables them to respond rapidly to the increased nutrient demand imposed by their increased growth rate when canopy openings occur.

The main objective was to evaluate the effects of seasonal variations in air temperature, soil water content, and irradiation on the net rate of CO₂ uptake for Agave tequilana growing in a warm subtropical environment (Amatitan, Jalisco, Mexico) and in a temperate subtropical environment (Arandas, Jalisco, Mexico). At Amatitan, the lowest value of positive daily net CO₂ uptake (141 mmol/m²/d) occurred at the end of the summer rainy season and the highest values during the winter (578 and 921 mmol/m²/d); at Arandas, the lowest value (188 mmol/m²/d) occurred at midsummer and the highest values at the end of the summer (763 mmol/m²/d) and in late winter (572 mmol/m²/d). The highest values of daily net CO₂ uptake in both localities reflected prolonged daily periods of net CO₂ uptake that included both daytime and nighttime assimilation accompanied by high rates of nocturnal net CO₂ uptake due to moderate nighttime air temperatures. High temperatures in the summer reduced daily net CO₂ uptake at both localities but in late summer, the nighttime temperatures at Arandas were moderate, which lead to high daily net CO₂ uptake. Leaf succulence allowed substantial daily net CO₂ uptake when the soil water content was relatively low during the winter and the spring (January–March). Indeed, appreciable daily net CO₂ uptake occurred throughout the year for A. tequilana at both localities, indicating that this CAM plant species can sequester carbon even during prolonged dry periods.

Most montane rain forests on the island of Hawaii consist of a closed canopy formed by Cibotium spp. tree ferns beneath an open canopy of emergent Metrosideros polymorpha trees. We used artificial seedlings to assess the extent to which physical disturbance caused by the senescing fronds of tree ferns and the activities of feral pigs might limit tree regeneration. Artificial seedlings were established terrestrially (N = 300) or epiphytically (N = 300) on tree fern stems. Half of the seedlings on each substrate were in an exclosure lacking feral pigs and half were in forest with pigs present. After one year, the percentage of seedlings damaged was significantly greater among terrestrial seedlings (25.7%) than epiphytic seedlings (11.3%). Significantly more terrestrial seedlings were damaged in the presence of pigs (31.3%) than in the absence of pigs (20.0%). Senescing fronds of tree ferns were responsible for 60.3 percent of the damaged seedlings. Physical disturbance is potentially a major cause of seedling mortality and may reduce the expected half-life of a seedling cohort to less than two years.

Liana cutting is a commonly suggested silvicultural practice aimed at reducing the negative impacts of lianas on timber production, but few experimental studies have been conducted to evaluate the cost and efficiency of this practice. In this study, we estimated the cost of cutting lianas in 12 plots of 0.25 ha each in a densely liana-infested forest of lowland Bolivia, and evaluated the efficiency of this silvicultural treatment in terms of the proportion of lianas missed, the density of resprouting liana stumps, and the number of liana-infested trees after two years of an experimental liana treatment.

The cost of cutting lianas in this forest by locally hired laborers was 23.6 (SE = 2.48) person-hours/ha. Considering local cost of labor and the U.S.–Bolivian currency exchange rate at the time of the study, this figure translates to ca $15/ha. Liana density decreased from 2471 (SE = 104.3) to 130 (SE = 24.2) liana stems ≥2 cm/ha immediately after cutting, because 5.5 percent of lianas were left uncut (missed). Slender lianas were missed more often than lianas with large-diameter stems. Liana species that grow 2–3 m before they start to twine were also frequently missed. Twenty-two percent of liana stumps ≥2 cm sprouted after cutting. Liana stumps with larger diameters sprouted more than stumps with smaller diameters. Most liana stumps produced only two sprouts. Two years after cutting, 78 percent of trees had no living lianas in their crowns, in contrast to only 13 percent liana-free trees in the control plots. Sixty-four percent of trees still had hanging dead lianas two years after cutting, but only 23 percent of trees were reinvaded by lianas using dead liana stems as trellises.

Liana cutting can efficiently reduce the number of lianas in liana-infested forests, and the effects of cutting lianas last for at least two years; however, the treatment is expensive. Thus, we recommend that it is better to view liana cutting as a preventive activity to avoid liana infestation, rather than as a corrective measure after poor management. Liana cutting can be easily conducted along with other reduced-impact logging practices.

Hurricanes have been assumed to reduce the reproduction of plants, either directly by leaf stripping and stress or indirectly by reducing pollinators. I examined the pollination and fruit set of a common shrub, Bourreria succulenta, after hurricanes on San Salvador island, Bahamas. Contrary to the assumption of resource limitation, B. succulenta showed unusually prolific flowering after Hurricane Lili stripped leaves from most of the plants in October 1996. I predicted that the abundant flowering would saturate pollinators and that fruit set would be pollination-limited. Fruit set was strongly pollination-limited by 71 percent. Butterflies are probably the major pollinators and were present at the site, but they rarely visited B. succulenta flowers even though flowers were brimming with nectar. Nectarivorous birds (Bananaquits and Bahama Woodstars) visit B. succulenta flowers, but their populations were decimated by Hurricane Lili and they rarely visited flowers during this time. Fruit set was also severely predation-limited; a moth caterpillar (Gelechiidae) was extremely abundant and ate buds, flowers, and fruits, causing a further 68 percent reduction in fruit set. Together, pollination limitation and predation limitation reduced fruit set to only 7 percent or less. Predation was also intense in 1999 after Hurricane Floyd and resulted in 11 percent fruit set or less. Whether or not hurricanes were the cause of limited pollinators or abundant predators, the resulting low fruit set could have population effects because hurricanes can provide opportunities for the recruitment of new plants. These results emphasize that understanding plant–animal interactions may be necessary for predicting the effects of hurricanes on plant reproductive success, which may affect subsequent recruitment. Species on small islands like San Salvador (150 km²) with relatively few species may be especially vulnerable to environmental disturbances such as hurricanes.

The diet of dingoes (Canis familiaris dingo) in the Australian Wet Tropics was examined by analyzing 383 dingo scats collected throughout the region for the presence of mammal prey remains. The scats yielded 29 native and 4 introduced mammal prey species from 14 families. The most important species in terms of percentage occurrence in the scats were Melomys cervinipes (22.2%), Isoodon macrourus (17.0%), Perameles nasuta (12.5%), and Thylogale stigmatica (12.5%). The most important families were Muridae (37.1%), Peramelidae (29.5%), and Macropodidae (25.8%). Examination of small-scale habitat preferences revealed species that preferentially use the forest edge ranked significantly higher in the diet than those that do not, and species that are terrestrial ranked higher in the diet than those that are arboreal. Relative abundance was also a significant factor in the ranked dietary occurrence of each species, with abundant species ranked significantly higher than those that are less abundant. These results suggest that dingoes in the Australian Wet Tropics are opportunistic predators of a wide variety of mammal species, with abundant terrestrial and forest edge-dwelling taxa the most susceptible to predation.

Home range sizes and spatial overlap of four sympatric squirrel species were investigated in a lowland dipterocarp forest in Malaysia using a radio-tracking method. The population density of Callosciurus caniceps was highest and C. notatus was next highest, while C. nigrovittatus and Lariscus insignis were scarce. C. caniceps was larger than C. nigrovittatus and C. notatus while L. insignis was extremely small. For females, home range size was smaller in L. insignis than Callosciurus spp., which may support the body weight hypothesis: larger species have larger home ranges. Among the three Callosciurus species, female C. caniceps had the smallest home range. These differences were accounted for by habitat characteristics rather than by density or body weight; C. caniceps was dominant in bushy areas and used crowded small trees while C. notatus and C. nigrovittatus used large trees in the forest. In this study, home range size did not change seasonally; this differs from studies in temperate regions, possibly because food availability is much less variable among seasons in tropical rain forest. Home range overlap among heterospecific individuals was common but different species seemed to partition space by using different vertical levels of the forest. Consequently, the home range size and spatial overlap of sympatric squirrel species may be affected by habitat diversity in tropical rain forest.

Endophytes are microorganisms that live inside plant tissues without causing symptoms of disease. They are a largely unexplored component of biodiversity, especially in the tropics. In this study, leaves from two populations of Guarea guidonia trees (Meliaceae) in the Luquillo region, Puerto Rico, were screened for diversity and composition of endophytic fungal communities. A total of 268 leaf fragments from 14 trees were surveyed. Thirty-eight morphospecies of endophytes were found. Phomopsis, Colletotrichum, Xylaria, and Rhizoctonia-like fungi were the most abundant taxa. Over 95 percent of the leaf pieces had endophytes. Communities had a few abundant species and many species with few individuals. The fungal community from the Guarea population in a forest preserve was more diverse than that from a disturbed area. Fungal communities were stratified according to height within a tree, but no differences were found between blade, petiolule, and rachis. The data suggest that the smaller and the more scattered the plant fragments sampled, the higher the probability of approaching real diversity values of endophytic fungal communities.