Modeling the spatial structure of topical forests

The spatial structure of tropical forest stands under different management conditions was modeled as a series of different spatial point processes. Spatial patterns were first assessed by K-function analyses to help choose a point process appropriate for observed patterns. The homogenous Neyman–Scott process accurately described live tree distribution in clear cut areas, where tree patterns tended to be aggregated. Parameters were estimated by minimizing Diggle's modified least squares criterion, and goodness-of-fit was assessed by comparison to confidence envelopes constructed by Monte Carlo simulation. Parameter estimates can be interpreted to help understand the ecological processes influencing re-colonization of disturbed areas. The inhomogeneous Poisson process was investigated for simulating the spatial pattern of ingrowth trees in lower canopy strata. The intensity function of this process was inversely proportional to variables representing canopy density. As assessed by Monte Carlo generation of confidence envelopes, the inhomogeneous Poisson process successfully portrayed the influence of canopy structure on understory plant distribution in most stands. Tree mortality was modeled as a thinning process in which the probability of individual tree mortality was conditional on subject tree attributes and competitive environment. The thinning function took the form of a generalized linear model with a binomial error distribution and logit link function. In most stands, tree neighborhood variables were powerful predictors of mortality, but they were not important predictors in all plots. This suggests that the surrounding forest structure of a subject tree has considerable influence on its morality, but competition is not the sole cause of tree morality in tropical forests.     

Height–diameter models for tropical forests on Hainan Island in southern China

Measurements on 8352 felled trees from 31 plots in tropical forests on Hainan Island provided information to investigate the relationship between total tree height and diameter at 1.3 m (DBH). Out of 33 model forms either published or derivable from published forms, a modification of the exponential model published by Meyer (Meyer, H.A. A mathematical expression for height curves. J. For. 38 (1940) 415–420) was selected based on low bias and relatively good precision of its predictions on individual plots. Variations from plot-to-plot were significant. Therefore, a parameter prediction method based on predictor variables derived from all DBH values on a plot was developed in order to provide a method to localize the model. In a test with independent data from an additional 36 plots, the localized height-prediction model produced predictions with a pooled mean-squared error of 3.8 m², a pooled R² of 0.72, and an average bias of −0.2 m. Even though tremendous variability exists in tropical forests and modeling height–DBH relationships never provides high levels of precision, this model will provide a useful and economical alternative to attempting height measurements during inventories of these forests.     

Fine litterfall and nutrient dynamics during forest regrowth in the humid subtropics of north-eastern India

Fine litterfall and nutrient return patterns were studied in three subtropical humid forest stands (7-, 13- and 16-year old), regrowing after selective tree cutting in north-eastern India. The seasonality of fine litterfall was unimodal, with a peak during spring and a trough during rainy season in the forest regrowths of three different ages. The rate of fine litterfall increased with increasing basal area of the woody vegetation during forest regrowth. Leaf litter accounted for 83% of the total litterfall. N concentration was maximum during autumn and minimum during rainy season; nutrient concentrations were highest in the leaf litter. Seasonal variation in P concentration was small. Maximum and minimum input of N and P to the forest floor through fine litter coincided with the respective periods of litter production.     

The geometry of a constant buffer-loading design method for humid watersheds

Riparian buffer strips are used in forestry to protect streams from possible adverse effects of forest harvesting or other land uses. For any given stream reach, a buffer loading can be defined as the contributing watershed area per unit area of buffer. The study used a large (66 km²) mountain watershed as a prototype. To allow accurate computation this was divided into facets by forming a flow net to the maximum accuracy of the 1 : 25 000 contour coverage. With fixed width buffers, the buffer loading was both highly variable and also independent of the Strahler order of the stream. Thus, the rationale of having larger buffers on larger streams does not seem justified. The study considered a buffer-strip design in which each element of stream buffer had exactly the same ratio of upslope-to-buffer area, giving a constant buffer loading. Computation of the buffer for each facet used an iterative procedure to achieve a satisfactory shape and position of the buffer boundary within each facet. The method gave a much more substantial protection to (convergent) channel sources and less protection to divergent areas than a fixed-width buffer design. The buffers defined also were highly asymmetric and discontinuous. The buffers defined reflected the topography, and were strongly influenced by small facets close to the stream. In cases where flow lines run close to and approximately parallel to the streams, the buffers defined were also non-intuitive. The method is predicated on the subsurface hydrology flow paths being close to those given by surface flow lines and this is not always true. Of importance is the finding that, relative to the mean protection offered, fixed-width buffers tend to underprotect slope convergences at the heads of streams, and overprotect divergent areas found along streams of increasing order.     

Modifying the microclimate around young oaks through vegetation manipulation: Effects on seedling growth and branching

The objective of this research was to evaluate the influence of vegetation control on the microclimate (light, air temperature, vapor-pressure deficit (VPD)) and on the growth of young planted, sessile oak seedlings. Three types of vegetation control, creating a range of above-ground interaction intensity, were compared. In the open treatment, vegetation was chemically controlled and was maintained at a low height; in the closed treatment, vegetation was manually controlled and a few dominant individuals that overtopped the oak seedlings were cut; and in the sheath treatment, the vegetation in a 0.5-m diameter circle around each seedling was manually cut to a height equal to three quarters of the total seedling height. In all treatments, a strong reduction of incoming radiation was observed, as well as an increase in air temperature and VPD during the day, and a decrease in daily PET, compared to the reference located 4 m above the vegetation. However, the neighboring vegetation did not modify the microclimate around the seedling to a level that could have induced a significant reduction in seedling photosynthesis and, thus, in seedling growth. Seven years after planting, seedlings were shortest in the open treatment and tallest in the closed treatment (334, 372, 378 cm in the open, sheath and closed treatments, respectively). Seedlings in the open treatment allocated proportionally more biomass to diameter increment than to height growth, compared to seedlings in the sheath and closed treatments. The seedlings averaged 1.77, 1.10 and 1.00 forks in the open, sheath and closed treatments, respectively, and the average height of the lowest fork was 171, 206 and 226 cm in the same treatments. The reduction of the number of forks per seedling was not related to a reduction of the appearance of new forks, but rather to a reduction in the persistence of the existing forks. These changes in stem allometry and in fork development seemed to be related to the lateral shade afforded by the vegetation.     

Determination of the spatial distribution of trees from digital aerial photographs

This study examined the possibilities of using computerized digital aerial photograph interpretation in determining the spatial distribution of trees. The material of the study included eight mapped stands in the municipality of Hyytiälä (61°50′N and 24°18′E), in southern Finland. The aerial photographs used were taken in June 1995 at a scale of 1:5000. Two approaches for determining the spatial pattern of trees were used. Firstly, in the point-process based approach used in this study, the individual trees in the digital aerial photograph were segmented by a robust segmentation method, based on recognition of the pattern of tree crowns with sub-pixel accuracy. Secondly, the crown coverage was determined by region growing segmentation combined with active surface representation. The significance of the differences in the means of image coverage pattern indices between the various spatial distribution categories was tested with one-way variance analysis. Because the process misclassified clustered spatial patterns as regular patterns, and regular patterns as random patterns, the usability of digital aerial photographs seems to be limited for the point-process based determination of the spatial pattern of trees if the scale is 1:5000 or less. When image coverage pattern indices were applied, the differences in the means of the spatial distribution categories proved not to be clearly statistically significant due to the great variation within classes. However, interpretation of crown coverage could have applications in practical forestry due to the low resolution requirements for the images used.     

Mitigating fire risk to late-successional forest reserves on the east slope of the Washington Cascade Range, USA

A fire-risk model was developed using a stand-structure approach for the forests of the eastern slopes of the Washington Cascade Range, USA. The model was used to evaluate effects of seven landscape-scale silvicultural regimes on fire risk at two spatial scales: (1) the risk to the entire landscape; and (2) the risk to three reserve stands with stand structures associated with high conservation priorities (layered canopy, large trees, multiple species). A 1000 ha landscape was projected five decades for each management regime using an individual tree, distance-independent growth model. Results suggest that a variety of silvicultural approaches will reduce landscape fire risk; however, reserve stand fire risk is minimally decreased by thinning treatments to neighboring stands. Intensive fuel reduction through prescribed burning and selection of reserve stands in favorable topographic positions provide substantial fire risk reductions.     

Climatic factors controlling the productivity of Norway spruce: A model-based analysis

The process-based growth model, BIOMASS, was modified to incorporate low-temperature effects on photosynthetic production in Norway spruce (Picea abies) stands growing in northern Sweden. The low-temperature features incorporated in BIOMASS made it possible to simulate and estimate the reduction in photosynthetic rates caused by boreal conditions. The following four simulation-scenarios were used: (i) ‘potential' photosynthesis without boreal restrictions; (ii) reduction caused by a frozen soil; (iii) reduction caused by incomplete recovery of photosynthetic capacity during spring as a result of damage caused by low winter temperatures; and (iv) reduction as an effect of frost-induced autumn decline. Annual photosynthetic production (or gross primary production (GPP)) was simulated for three calendar years, 1990–1992, for stands with low (control) and high (irrigated and fertilized) nutrient availability. The reduction of ‘potential' GPP, caused by the low-temperature effects, ranged from 35–44% for control (C) and from 34–42% for irrigated-fertilised (IL) stands, respectively. The most pronounced loss of ‘potential' GPP originated from reduced photosynthetic capacity, in spring and early summer, which led to losses of 21–28% for C and 19–26% for IL stands. The variation between years differed mainly as an effect of differences in spring temperatures, which resulted in different rates of recovery of photosynthetic capacity. Reductions caused by frozen soil and low photosynthetic capacity during winter were similar in C and IL stands (12–13%), as were the losses resulting from severe autumn frosts (3–4%). It is concluded that, unless the effects of frozen soils and reduced photosynthetic capacity during spring and early summer are considered, large errors (ca. 40%) will be introduced into estimates of the annual photosynthetic production of boreal conifer forests.     

Testing site index–site-factor relationships for predicting Pinus contorta and Picea engelmannii×P. glauca productivity in central British Columbia, Canada

Measures of forest productivity for various site conditions are necessary for forest management planning, where timber production is the objective. This study was undertaken to test whether autecological productivity relationships developed for lodgepole pine (Pinus contorta) and interior spruce (Picea engelmannii×P. glauca) using the biogeoclimatic ecosystem classification system of British Columbia are useful as practical field-based procedures to predict site index. Independent data sets consisting of 111 plots for pine and 114 plots for spruce were collected for use in testing the bias and precision of the models. A regression on residuals (predicted minus test values) indicated that the lodgepole pine model was unbiased in estimating site index (p=0.08). However, there was a lack of precision, with a square root of the mean squared prediction error (root-MSPR) of 2.8 m. Only 56% of the test plots had differences from the predicted values of 2.0 m or less. Residual analysis showed that the interior spruce model was biased in estimating site index (p<0.01), generally predicting greater site index than the test values. The model also lacked precision, with a root-MSPR of 3.2 m. Only 44% of the test plots had differences from the predicted values of 2.0 m or less. Forest managers requiring a site-index prediction tool need to decide whether the degree of accuracy precision provided by these models is acceptable.     

Effects of age and frequency of cutting on productivity of Mediterranean deciduous fodder tree and shrub plantations

Deciduous fodder trees and shrubs are important feed resources for the critical summer period in semi-arid and sub-humid Mediterranean environments. Eleven such species were established in central Macedonia, Greece as 1-year old seedlings in 1987 and studied through 1994. Height and biomass were statistically different among species and tended to increase with the age of plants. Common Robinia pseudoacacia was the tallest and the most productive species, reaching more than 3 m height and 4 kg DM/plant 8 years since planting; it was followed by Ostrya carpinifolia, the variety monophylla of Robinia, Colutea arborescens and Amorpha fruticosa. Grazeable proportion of total biomass was about 50% and decreased as the age of plants increased. Repeated annual cutting resulted in significant reduction of both height and total biomass by 51% and 88%, respectively as compared with uncut plants at the end of the eighth year while frequent cutting tended to increase the grazeable proportion of biomass after the fourth year. To attain an optimum combination of quantity and quality of biomass in deciduous fodder tree and shrub plantations cutting or grazing should not start earlier than the third year since establishment.     

Physiological and growth responses of three sizes of containerized Picea mariana seedlings outplanted with and without vegetation control

Three different stock sizes of containerized black spruce (Picea mariana [Mill.] B.S.P.) seedlings were planted in an abandoned agricultural field. The small planting stock was of a conventional type produced in 110 cm³ containers. The experimental medium and large stock types were produced in 340 and 700 cm³ containers, respectively. Gas exchange, xylem water potential and dry masses were measured six times during each of the first two growing seasons in field plots with and without vegetation control. During the first growing season, the effect of planting shock masked most physiological and growth differences among seedling types. During the second growing season, in plots with vegetation control, small and medium seedlings had similar values of physiological variables and of growth as measured by relative growth rates (RGR), but the large seedlings showed lower values of both net photosynthesis and of RGR, a difference attributed to low initial quality of the root system in the larger seedlings. In plots without vegetation control, the trend was identical, but differences were not significant; the greater height of the larger seedlings, and the resulting greater access to light, compensated for their lower initial quality. The similarity in response between the medium and the small seedlings shows that a fourfold increase in shoot size (1.68–6.82 g) in the initial size and a doubling of the shoot : root ratio (2.17–4.54) of the planting stock did not result in increased planting shock or reduced growth in these containerized conifer stock types. The results also show the importance of the interaction between stock height and the vertical light profile created by the competing vegetation in the final assessment of stock performance.     

A comparison of plant and carabid beetle communities in an Irish oak woodland with a nearby conifer plantation and clearfelled site

In 1996, plant and carabid beetle communities were examined in a semi-natural oak woodland (Derryclare Wood) and compared with communities of a nearby mature conifer plantation and a plantation clearfelled in 1994. At each site, 10 quadrats (2 m×2 m) and 10 pitfall traps were used to sample the ground vegetation and carabid beetle communities, respectively. Forty-one plant species were recorded in the oak woodland while 38 species were found in the clearfelled site and 19 species in the conifer plantation. The median plant-species richness per quadrat was significantly greater (p<0.01) in the oak woodland (16.5) than in the other two sites and the clearfelled site (11) was significantly more species-rich (p<0.02) than the conifer plantation (6.5). A total of 21 carabid species (895 individuals) were captured, with 74.5, 13.9 and 11.6% of individuals being caught at the oak woodland, clearfelled site and conifer plantation, respectively. The median number of carabid species per pitfall trap was significantly greater (p<0.01) in the oak woodland (6) than in the clearfelled site (3.5) and conifer plantation (2). In addition, median species richness per pitfall trap was significantly higher in the clearfelled site than in the conifer plantation at p<0.02. Median diversity per pitfall trap was also significantly less in the conifer plantation (1.19) than in the oak woodland (2.43, p<0.01) and clearfelled site (1.55, p<0.05). While the Sorenson similarity index for both vegetation and carabids shows that the conifer plantation and the clearfelled site were most similar in species composition; nevertheless, the clearfelled site has demonstrated signs of increasing species richness only 2 years after clearfelling. It is concluded that while semi-natural woodlands support a greater range of species than conifer plantations, clearfelled areas which are not replanted can enhance species richness within commercial conifer plantations.     

Phosphorus and potassium depletion by roots of field-grown slash pine: Aerobic and hypoxic conditions

Slash pine (Pinus elliottii Engelm. var. elliottii) is planted extensively on flatwood and coastal savanna soils of the southeastern USA. Pine roots growing in these soils encounter shallow water tables. Although the fine-root system of pine trees growing in the surface 20–26 cm of those soils will not be continuously submerged, they will encounter short-term reduced soil conditions, with a severely reduced O₂ supply which might affect their ability for ion uptake from the soil solution. The objectives were: (i) to compare P and K depletion by lateral root systems of slash pine roots subjected to a short-term hypoxic treatment, and (ii) to document K and P depletion patterns by lateral root systems of slash pine roots following the removal of the hypoxic treatment. Our purpose was to evaluate the uptake ability of these roots of changing aeration. For the experiment, 17 intact lateral roots from twelve-year-old slash pine trees were inserted into nutrient-uptake root chambers. The chambers were filled with a nutrient solution containing 6.25 μM P (phosphate), and 25.66 μM K. P and K depletion rates were monitored in six consecutive treatments, each lasting five days. Treatments of aerobic and hypoxic conditions were sequentially applied in the following order: aerobic–hypoxic–aerobic. This sequence was repeated twice. Uptake of P and K by slash pine roots was affected by oxygen availability, but the degree of response differed. Under hypoxic nutrient solution conditions, K depletion from solution by pine roots was totally inhibited, resulting in net efflux of K. In contrast, P depletion was not inhibited under hypoxic nutrient solution conditions. Results suggest that pine roots grown in aerobic soil conditions of surface horizons are capable of P depletion when reduced soil conditions are present.     

The impact of aspect on forest structure and floristics in some Eastern Australian sites

Aspect exerts a major influence on the structure and floristics of forests at upslope positions on steep ridges near the escarpment of north-eastern New South Wales. The vegetation of northwest facing sites differed from that of the southeast aspect in species richness and diversity, canopy cover, and canopy height, whilst differences between northeast and southwest aspects were significant for fewer attributes. The gross effect of aspect was in segregating eucalypt and rainforest communities. Aspect-induced differences in light, heat, moisture and edaphic conditions produced a sort of potential canopy species during recruitment. In the absence of substantial disturbance, the divergence of the structural character of developing forest canopies between aspects tends to enhance the floristic divergence in the subcanopy space. Although the study area is representative of a large region of forested steeplands in the hinterland of north-eastern New South Wales, the upper slope segments and four aspects studied can be expected to represent the extreme manipulation. In any event, it is unlikely that the extent of site divergence represented in these old-growth forests would be maintained under managed re-growth strategies.     

Patterns of forest use and endemism in resident bird communities of north-central Michoacán, Mexico

We compared breeding avian communities among 11 habitat types in north-central Michoacán, Mexico, to determine patterns of forest use by endemic and nonendemic resident species. Point counts of birds and vegetation measurements were conducted at 124 sampling localities from May through July, in 1994 and 1995. Six native forest types sampled were pine, pine–oak, oak–pine, oak, fir, and cloud forests; three habitat types were plantations of Eucalyptus, pine, and mixed species; and the remaining two habitats were shrublands and pastures. Pastures had lower bird-species richness and abundance than pine, oak–pine, and mixed-species plantations. Pine forests had greater bird abundance and species richness than oak forests and shrublands. Species richness and abundance of endemics were greatest in fir forests, followed by cloud forests. Bird abundance and richness significantly increased with greater tree-layer complexity, although sites with intermediate tree complexity also supported high abundances. When detrended correspondence-analysis scores were plotted for each site, bird species composition did not differ substantially among the four native oak-and-pine forest types, but cloud and fir forests, Eucalyptus plantations, and mixed-species plantations formed relatively distinct groups. Plantations supported a mixture of species found in native forests, shrublands, and pastures. Pastures and shrublands shared many species in common, varied greatly among sites in bird-species composition, and contained more species specific to these habitats than did forest types.     

Phosphorus cycling and soil P fractions in Douglas-fir and red alder stands

Nitrogen-fixing species can dramatically increase soil acidity and organic matter content, and potentially alter biogeochemical P dynamics. We compared ecosystem P cycling in adjacent stands of N₂-fixing red alder (Alnus rubra Bong.) and non-fixing Douglas-fir (Pseudotsuga menziesii Franco) in order to determine whether P-cycling rates within stands were related to soil P forms as measured by sequential P fractionation. Above-ground annual P uptake was 61% greater in the red alder stand, although soil available P, as measured by Bray (NH₄F–HCl) extraction, was only 10% of that found in the Douglas-fir stand. Total ecosystem P in the alder stand was only 69% of that found in the Douglas-fir stand, and could indicate a pre-establishment difference between stands. However, the percentage of total soil P released by Bray or NaOH extraction was also lower in the alder stand, which suggests that differences in total P alone did not control the patterns observed in P fractions. Concentrations of inorganic P sorbed to Fe and Al minerals and contained in Fe minerals and apatite were greater under Douglas-fir, while organic P was slightly greater under red alder. While fluxes of P in litterfall, uptake and resorption were 94, 60 and 292% higher in the alder stand, soil extractable fractions meant to represent available P were lower under alder. Static measures of available P do not appear to adequately reflect P supply, and the development of techniques to assess P turnover is needed to better understand cycling and plant availability of P.     

Genetic control of growth and form in early-age tests of Casuarina equisetifolia in Andhra Pradesh, India

Early results from three test plantings of Casuarina equisetifolia spp. equisetifolia are reported: a clonal test of 40 locally selected clones, a progeny test of 16 wind-pollinated families of local origin, and a family-in-provenance test of 100 wind-pollinated families which is part of an international provenance trial. All tests were planted in 1996 at the Regional Forest Research Centre, Rajahmundry, Andhra Pradesh, India. Height, stem straightness and axis persistence were measured in the family tests at 6 and 12 months of age. The height of the clonal test was measured at 12 months of age. Heritability estimates obtained were ca. 0.3, 0.2 and 0.1 for height, stem straightness and axis persistence, respectively. Heritability estimates for height were reasonably consistent across the three tests; however, estimates for stem straightness and axis persistence were generally higher in the small test of locally selected families. Height and the other two traits are most probably either weakly correlated or uncorrelated, while stem straightness and axis persistence are positively correlated. Provenances spread across the range of natural- and land-race populations tested demonstrated good growth potential at Rajahmundry. However, natural provenances from Malaysia and Thailand generally showed superior growth, while the land-race populations often demonstrated superior stem straightness. Based on these results some aspects of a breeding strategy are outlined: (a) the establishment of three unrelated sublines; (b) an initial clonal test of all parents; and (c) wind-pollinated mating within clonal breeding orchards.     

Structure of silver fir (Abies alba Mill.) natural regeneration in the ‘Jata'' reserve in Poland

Over the last 20 years, investigations have been carried out to determine the influence of various ecological factors on silver fir natural regeneration in highlands and mountains. However, there has been little research on the structure and development of fir regeneration in lowlands. Results of this study indicate that three main stand characteristics play a very important part in the structure of fir regeneration in the lowland. The results revealed that the quantity, frequency and growth rate of fir regeneration were affected by site conditions. One of the most important ecological factors differentiating quantity and quality of fir regeneration was the proportion of fir in a stand. It was found that, with an increase in the percentage of fir in a stand, the quantity and the sum of heights and the sum of height increments of fir regeneration tends to increase. Results of this study showed that the number and development of fir regeneration were influenced by species composition of a stand; fir regenerated not only in pure fir stands but also in mixed forests. A positive influence of pine and birch canopy on initiation and development of fir regeneration was confirmed. Optimal conditions for the growth and development of fir with respect to species composition were found in mixed fir stands with an admixture of hornbeam. In contrast, results of the study suggest that the worst conditions for fir regeneration were found in the stands composed of species, such as ash, alder, oak, aspen, lime and spruce.     

Influence of artificial gaps in tropical forest on survival, growth, and Phytolyma lata attack on Milicia excelsa

Milicia excelsa and M. regia are important timber species in moist tropical areas of Africa. They have not been successfully grown largely because of attacks by gall-forming psyllids in the genus Phytolyma. Our objectives were to evaluate the growth of planted Milicia seedlings and incidence of psyllid attack in small (4.2 m²), medium (18.5 m²) and large (>500 m²) artificial gaps in the Bobiri Forest Reserve in the Moist Semi-Deciduous Forest of Ghana. After 13 months, height and diameter growth of Milicia seedlings were significantly greater in the medium and large gaps than in the small gaps. Insect attacks occurred first and most severely in the large gaps, but spread to gaps of all sizes between the 11th and 13th months after planting. While gap size significantly affected the susceptibility of seedlings to psyllid attack, it is not the only factor important in determining susceptibility of Milicia excelsa under field conditions. We conclude that gap sizes in the range of 10–50 m², where irradiances are from 30–60% of full sunlight, in forests similar to those at the study site, seem to be most suitable for regeneration of Milicia.     

Tropical forest fragmentation and greenhouse gas emissions

Rainforest fragments in central Amazonia have been found to experience a marked loss of above-ground biomass caused by sharply increased rates of tree mortality and damage near fragment margins. These findings suggest that fragmentation of tropical forests is likely to increase emissions of CO₂ and other greenhouse gases above and beyond that caused by deforestation per se. We estimated committed carbon emissions from deforestation and fragmentation in Amazonia, using three simulated models of landscape change: a ‘Rondônia scenario,' which mimicked settlement schemes of small farmers in the southern Amazon; a ‘Pará scenario,' which imitated large cattle ranches in the eastern Amazon; and a ‘random scenario,' in which forest tracts were cleared randomly. Estimates of carbon emissions for specific landscapes were from 0.3 to 42% too low, depending on the amount and spatial pattern of clearing, when based solely on deforestation. Because they created irregular habitat edges or many forest perforations which increased tree mortality, the Rondônia and random-clearing scenarios produced 2–5 times more fragmentation-induced carbon emissions than did the Pará scenario, for any given level of clearing. Using current estimates of forest conversion, our simulations suggest that committed carbon emissions from forest fragmentation alone will range from 3.0 to 15.6 million t/year in the Brazilian Amazon, and from 22 to 149 million t/year for tropical forests globally.