Abstract
Trees structure the Earth’s most biodiverse ecosystem, tropical forests. The vast number of tree species presents a formidable challenge to understanding these forests, including their response to environmental change, as very little is known about most tropical tree species. A focus on the common species may circumvent this challenge. Here we investigate abundance patterns of common tree species using inventory data on 1,003,805 trees with trunk diameters of at least 10 cm across 1,568 locations1,2,3,4,5,6 in closed-canopy, structurally intact old-growth tropical forests in Africa, Amazonia and Southeast Asia. We estimate that 2.2%, 2.2% and 2.3% of species comprise 50% of the tropical trees in these regions, respectively. Extrapolating across all closed-canopy tropical forests, we estimate that just 1,053 species comprise half of Earth’s 800 billion tropical trees with trunk diameters of at least 10 cm. Despite differing biogeographic, climatic and anthropogenic histories7, we find notably consistent patterns of common species and species abundance distributions across the continents. This suggests that fundamental mechanisms of tree community assembly may apply to all tropical forests. Resampling analyses show that the most common species are likely to belong to a manageable list of known species, enabling targeted efforts to understand their ecology. Although they do not detract from the importance of rare species, our results open new opportunities to understand the world’s most diverse forests, including modelling their response to environmental change, by focusing on the common species that constitute the majority of their trees.
Similar content being viewed by others
Main
Tropical forests are a crucial component of the Earth system; they cover around 10% of the Earth’s land surface8 but contribute approximately 33% of terrestrial net primary productivity9. They account for around 40% of the carbon stored in live vegetation10 and are globally important carbon sinks11. Tropical forests are also extraordinarily biodiverse, harbouring two-thirds of all known species12 and the majority of the world’s biodiversity hotspots13. Of note, as many tree species can be found in a single hectare of tropical forest as in the entire native Western European tree flora14. Recent estimates suggest that there are approximately 37,900 named tropical tree species in the scientific literature15, with potentially thousands more yet to be identified by scientists16. This extraordinary diversity means that little is known about the biology of the vast majority of tropical tree species. Our understanding of tropical forest ecology, productivity and carbon storage and how they may respond to environmental change is hindered by this lack of knowledge. This limited understanding also curtails scientific input into land use, biodiversity, climate and other forest-related policy and management.
Our understanding of tropical forests may improve through a focus on the most common tree species. This is a promising avenue, given that species abundance distributions (SADs) showing a modest number of common species and much larger numbers of rare species have been documented across taxa globally17,18,19. Indeed, analyses of tropical forest inventory data from Amazonia have shown that a relatively small number of common species comprise a majority of trees in the region6,20,21,22,23,24. However, whether such patterns hold in other tropical forests is unknown, as there have been no comparable analyses for African or Southeast Asian tropical forests. Perhaps, given the substantial differences in total tree species richness25, forest structure1, contemporary climate26 and biogeographic and human-occupancy histories7 among continents, important contrasts in patterns of common species would be expected. Alternatively, if the same processes or mechanisms apply to all tropical forests27, highly consistent patterns may be expected. Crucially, if a tractably modest number of common species do comprise the majority of tropical trees on Earth, this could open new ways of understanding tropical forests by investigating the ecology of the common species.
Cross-continental comparisons of common species patterns are complicated by unresolved differences in the results from published Amazon forest studies6,20,22. Estimates of hyperdominance—describing the minimum number of species required to account for 50% of all trees in a sample—range from 1.4% to 8.2% of the total number of species found in each of the Amazon forest datasets analysed (corresponding to 224 and 1,312 hyperdominant species respectively, assuming 16,000 Amazon tree species). Therefore, here we: (1) investigate sample-related biases and standardize our sampling to enable meaningful comparisons among datasets; (2) test whether patterns of hyperdominance differ across Amazonia, Africa and Southeast Asia; (3) extrapolate our results to assess how many species comprise half of all Earth’s tropical trees; (4) assess species abundance patterns, with differing classifications of ‘common species’ beyond hyperdominance; and (5) use resampling techniques to assess which sampled species are likely to be hyperdominant.
We analyse species abundance data from networks of inventory plots across three continents. We limit our analysis to closed canopy structurally intact old-growth tropical forests. For Amazonia, defined as the lowland Amazon Basin and Guiana Shield, we use the Amazon Tree Diversity Network and RAINFOR datasets (n = 1,097 plots). For Africa, encompassing West, central and East Africa, we use the African Tropical Rainforest Observatory Network (AfriTRON)1, Central African Plot Network, and two smaller networks2,3 (n = 368 plots). For Southeast Asia, defined as extending from Myanmar in the West to Sulawesi in the East, we use a tree diversity4 and a carbon monitoring5 network (n = 103 plots). We limit our analysis to trees with trunk diameter of at least 10 cm at breast height (1.3 m along the stem or above any buttresses or deformities), the widely used minimum size for inventorying tropical trees. The combined dataset includes 1,003,805, trees, of which 93.3% are identified to species (Fig. 1 and Extended Data Table 1).
Dots show the location of the plots analysed, coloured by continental region. Dark green shows the Amazonia, Africa and Southeast Asia regions that we extrapolate to. Light green shows ‘tropical and subtropical moist broadleaf forests’60, which we extrapolate to as the closed canopy tropical forest biome.
Consistent patterns of commonness
The Africa, Amazonia and Southeast Asia datasets differ in the number and size of plots sampled and the number of trees sampled (Extended Data Table 1). We therefore excluded small plots (below 0.9 ha; Extended Data Fig. 1 and Methods) and used rarefaction—that is, repeated random subsampling of plots to comparable numbers of trees—to standardize sampling across the three datasets (Fig. 2).
a–d, The effect of increasing sample size on the number of hyperdominants (a), total species (b), hyperdominant percentage (c) and fitted values of Fisher’s α (d) in tropical Africa (magenta), Amazonia (cyan), Southeast Asia (blue). Rarefied data (mean values across iterations of subsamples) are shown as points joined by lines for clarity, shaded areas represent 95% confidence intervals (derived via the s.d. across iterations of subsamples taken with replacement at each sampling point). Note that resampling for rarefaction was by subsampling of plots, but curves are re-plotted on an x axis of number of stems.
Rarefying to a common sample size of 77,587 stems, the size of the Asia dataset (equivalent to 150, 116 and 103 plots in Africa, Amazonia and Southeast Asia respectively), we find that 77 species (95% confidence interval: 62–92) in Africa comprise 50% of individual trees, compared with 174 species (95% confidence interval: 134–215) in Amazonia and 172 species (95% confidence interval: 125–217) in Southeast Asia (Table 1 and Fig. 2). However, the substantially lower number of hyperdominant species in Africa compared with Amazonia and Southeast Asia scales with the substantially lower number of total species. We find just 1,132 species in our standardized 77,587 tree sample in Africa, compared with 2,565 and 2,585 species in Amazonia and Southeast Asia, respectively for the same sample size. Consequently, percentage hyperdominance is statistically indistinguishable among the continents at 6.79% (95% confidence interval: 5.39%–8.20%), 6.80% (95% confidence interval: 5.24%–8.36%) and 6.65% (95% confidence interval: 4.59%–8.71%) in Africa, Amazonia and Southeast Asia, respectively (Table 1). This consistency is not affected by the aggregated spatial distribution of plots within each region (Extended Data Fig. 2) and holds true for analyses based solely on 1-ha plots (Methods). Thus, once sampling is standardized, there is marked pan-tropical consistency in the proportion of the total number of tree species accounted for by the most common species.
The consistency of commonness is not limited to defining common species as those that account for 50% of all individual trees in a dataset. The proportions of the total number of species required to account for thresholds between 10% and 90% of individual trees are also highly consistent across the rarefied data for the three continents (Fig. 3 and Extended Data Table 3). Thus, the data from the three continents appear to result from the same underlying statistical distribution.
Circles show results as rarefied to the size of the Southeast Asia dataset (mean values across iterations of subsamples with 77,587 stems). Diamonds show the extrapolated results at the scale of the regions. Estimated rarefaction confidence intervals are derived from the s.d. across iterations of subsamples taken with replacement at 77,587 stems.
Our rarefaction analysis shows that the number of hyperdominants, the total number of species and the percentage hyperdominance are dependent on sample size. This is because as plots—and therefore trees—are added to the sample, increasing numbers of rare species start to appear. Meanwhile, most common species have, by definition, already appeared, but their abundances increase. Thus, with increasing sample size, the number of hyperdominants increases, but at an ever-decreasing rate that tends towards saturation (Fig. 2 and Extended Data Fig. 3). The total number of species increases at a decreasing rate with increasing sample size, without apparent saturation. Therefore, as sample sizes increase, the percentage hyperdominance decreases gradually, but does not appear to saturate (Fig. 2 and Extended Data Fig. 3). This sample size dependence is likely to explain the published differences in percentage hyperdominance in Amazonian forests, which follow expectations given the sample size in each study6,20,22.
Amazonia and Southeast Asia show remarkably similar patterns of commonness and diversity. The rarefaction curves of the number of species accounting for 50% of all trees (Fig. 2a), total number of species (Fig. 2b), percentage hyperdominance (Fig. 2c) and Fisher’s α—the parameter of the log series distribution shown to best describe tropical tree species abundance distributions21 (Fig. 2d)—are almost identical between the two datasets. Furthermore, the numbers of species required to account for any threshold between 10% and 90% of trees in the respective rarefied samples of 77,587 trees are statistically indistinguishable (Table 1 and Extended Data Tables 2 and 3). This equivalence in overall tropical forest diversity patterns between these similarly species-rich regions is particularly striking given their very different biogeographic, climatic and anthropogenic histories, and the fact that Amazonia is one large contiguous region, whereas Southeast Asia is a series of islands and island-like regions.
In contrast to the similarity between Amazonia and Southeast Asia, our results provide sample size-corrected validation of the ‘odd-one-out’ observation28,29 of much lower tree species richness in Africa compared with Amazonia and Southeast Asia. Here we add a similar odd-one-out observation of a much lower number of common species in Africa than in Amazonia and Southeast Asia. However, in combination these two results lead to an almost identical percentage hyperdominance in the African, Amazonian and Southeast Asian rarefied data. This consistency extends to the proportion of species required to account for all thresholds between 10% and 90% of trees in the rarefied data (Fig. 3 and Extended Data Table 3). This pan-tropical invariance recasts the tropical forests of Africa from ‘odd’ in terms of species richness to statistically indistinguishable from those in Amazonia and Southeast Asia in terms of proportional patterns of abundance. Overall, using standardization by rarefaction, we find consistent patterns of species abundance across Africa, Amazonia and Southeast Asia.
Scaling to the study region
Next, we estimate commonness patterns in each of our three study regions: Africa, Amazonia and Southeast Asia. We extrapolate log series fits to the empirical Africa, Amazonia and Southeast Asia datasets (Extended Data Fig. 4), including a correction to account for the clumped spatial occurrence of species, to the total number of trees with trunk diameter of at least 10 cm in each study region. We estimate that just 104 species (95% confidence interval: 101–107) account for 50% of the 113 billion trees in Africa’s closed canopy tropical forests (Table 2). We also estimate that just 299 species (95% confidence interval: 295–304) account for 50% of the 344 billion trees in Amazonia’s closed canopy tropical forest, and 278 (95% confidence interval: 268–289) account for 50% of the 129 billion trees in Southeast Asia’s closed canopy tropical forests (Table 2). Our results from Amazonia match those derived using a different extrapolation approach30.
Our extrapolations again outline consistent percentage hyperdominance: just 2.2% of African, 2.2% Amazonian and 2.3% of Southeast Asian species account for 50% of all trees with trunk diameters of at least 10 cm in each region (Table 2). The dominant proportions of total species required to account for 10% to 90% of trees are also very similar across continents (Fig. 3 and Extended Data Table 5). The lower percentage dominance values from the extrapolated data compared with those from the rarefied data are consistent with the pattern, described above, of many more rare species being added as the number of trees increases while many fewer common species are added (Fig. 2). Overall, the extrapolated results show that there are a tractable number of common species in tropical forests in Africa, Amazonia and Southeast Asia.
Scaling to the tropics
We next estimate the number of common tropical tree species on Earth by multiplying the pan-tropical proportion of common species by the total number of tropical tree species on Earth. Our results suggest a pan-tropical hyperdominant percentage of 2.24% (Table 2). However, our extrapolations cannot provide an estimate of the total number of tropical tree species because we do not—for this study—have data from all tropical regions, including a lack of data from Central America, New Guinea and Micronesia. Furthermore, there is no consensus estimate of the total number of tropical tree species on Earth.
A compilation of lists of species known to science suggests a total of 60,065 tree species globally15. Tropical forest biomes likely comprise 63% of this list (E. Beech, personal communication, 2021), implying that there are around 37,900 known tropical tree species. This minimum estimate does not account for species that are yet to be identified and described by scientists. An alternative extrapolation method estimated that there are 46,900 species for the closed canopy tropical forest biome25 (range 40,500–53,300 species), implying that there are 9,000 yet-to-be-identified species. This is in agreement with a recent global study suggesting that there are around 9,200 tree species remaining yet to be formally named, almost all in the tropics16. Thus, together, these studies suggest there are likely to be approximately 47,000 tropical tree species in the world’s closed canopy tropical forests.
Our best estimate is that 1,053 tree species (2.24% of 47,000 species) account for half of Earth’s 800 billion trees with trunk diameters of at least 10 cm found in the closed-canopy tropical forest biome. Although the true number may be lower or higher, the conclusion that a tractable number of species dominate tropical forests is clear. Some of these species are likely to be extraordinarily common: our best estimate is that just 61 species account for 80 billion individual trees (0.13% of 47,000 species). At the other end of the spectrum, we estimate that the rarest approximately 39,500 species account for just 80 billion trees, or 10% of individuals. Meanwhile, the other 90% of all trees are estimated to belong to just 7,487 species (15.93% of 47,000 species). Thus, these results open the possibility of focusing efforts on understanding the biology of a tractable number of species in tropical forests to approximate the whole stand.
Identifying the most common species
Our analyses showing that 104, 299 and 278 common species account for 50% of the trees in our African, Amazonian and Southeast Asian study regions, respectively, do not yield a list of named species. To assess which named species are likely to be hyperdominant, we use a subsampling procedure similar to the rarefaction methodology above. We randomly subsample from approximately 10,000 trees per subsample (drawn by plot) and increase the size of the subsample in 10,000-tree increments until the size of each regional dataset is reached, and repeat this process 100 times. For each sampled increment of 10,000 trees we then calculate the proportion of random subsamples in which each species qualifies as hyperdominant (Supplementary Table 1). We then assign the species to one of four groups:
-
(1)
Both hyperdominant in the full data and hyperdominant in the majority of subsamples even at very small sample sizes. These 50, 95 and 105 species in our Africa, Amazonia and Southeast Asia datasets, respectively, represent 3.5%, 2.1% and 4.1% of sampled species in each dataset. These species are likely to be geographically widespread and abundant.
-
(2)
Both hyperdominant in the full data and hyperdominant in the majority of subsamples, but at the smallest sample sizes only occasionally hyperdominant. These 32, 129 and 67 species in our Africa, Amazonia and Southeast Asia datasets, respectively, represent 2.3%, 2.9% and 2.6% of sampled species in each dataset. These species are likely to be geographically widespread but not always abundant.
-
(3)
Not quite hyperdominant in the full data, but hyperdominant in a substantial proportion of subsamples. These 102, 339 and 200 species in our Africa, Amazonia and Southeast Asia datasets, respectively, represent 7.2%, 7.5% and 7.7% of sampled species in each dataset. These species are probably locally abundant but not necessarily geographically widespread.
-
(4)
Not hyperdominant in the full data and almost never hyperdominant in the subsamples. These 1,232, 3,929 and 2,213 species in our Africa, Amazonia and Southeast Asia datasets, respectively, represent 87%, 87.5% and 85.6% of sampled species in each dataset. These species are probably neither geographically widespread nor abundant.
We suggest that if all trees in a region were sampled, the hyperdominant species would be drawn from the first three groups, which are listed in Supplementary Table 2. This candidate list of 1,119 hyperdominant species contains 184 species in Africa, 563 species in Amazonia and 372 species in Southeast Asia, with no species appearing on more than one region’s list. Thus, the list of species that are likely candidates for hyperdominance is manageably small.
There is uncertainty in our candidate hyperdominant list owing to the limitations of the underlying samples of plots across the landscape. Specifically, some species that always have low local abundance but are geographically widespread and lack habitat restrictions may require larger sample sizes for their hyperdominance to become clear. Similarly, species that combine low local abundance and habitat specificity pose challenges. If the distribution and extent of specialist habitat is great enough to result in hyperdominance of specialists but is not sufficiently captured in our sampling, such species might not appear in our candidate list. By contrast, some species in our candidate hyperdominant list will not be true hyperdominants. Of particular note, some apparently common species may actually comprise a group of cryptic species, with none of these cryptic species being hyperdominant by itself31,32,33. However, the striking similarly in species abundance patterns across the Africa, Amazonia and Southeast Asia datasets, despite differing sampling intensity on each continent, suggests that these potential limitations do not substantially affect the overall patterns found. We therefore expect a high overlap between our list of candidate hyperdominant species and eventual elucidation of the actual hyperdominants of these three regions and the pan-tropics.
Our list of 1,119 candidate hyperdominant species represents a tractable number of species on which to prioritize autecological research. Indeed, given their commonness, ecological data already exists for many of these species: 95% have some autecological data recorded in a large global database34; 83% have at least 10 different types of measurement, typically including their growth form, maximum height, wood density and aspects of leaf chemistry. This indicates that these species are already relatively well known. Therefore, only limited additional data may be required to open new approaches to better understanding tropical forests through their most common tree species, including how they may react to today’s era of rapid global environmental change.
Discussion
Charles Darwin wrote in The Origin of Species that “rarity is the attribute of a vast number of species of all classes and in all countries”35. If this is the case, then common species are themselves rare. Our results concur: despite their formidable diversity, the trees in tropical forests fit the ‘rare is common, common is rare’ pattern36 which has been documented in many other taxa17,18,19,36,37. Beyond this, our analyses reveal highly consistent patterns of commonness across three major tropical forest regions. Notably, despite substantial inter-continental variation in biogeographic history, contemporary environment, forest structure and species composition, we have found an emergent property of the tropical forest system. For the trees that structure tropical forests, a consistent ~2.2% of the total species pool accounts for 50% of all individual trees in Africa, Amazonia and Southeast Asia. This consistency is all the more notable given relatively lower tree species richness of African tropical forests compared with Amazonian and Southeast Asian forests, probably owing to higher extinction rates in African forests, with evidence of major losses of African species at the Oligocene–Miocene boundary38, and contractions of rainforest area due to drier conditions during repeated glacial–interglacial cycles over the past 2.6 million years39.
We find common diversity patterns despite the very different histories of human occupancy in Amazonian, African and Southeast Asian tropical forests40. The relatively recent arrival of humans in Amazonia approximately 20,000 years ago has been linked to greater Pleistocene extinctions, in contrast to much longer human occupancy in the tropical forests of Africa and Southeast Asia41. Some have also suggested that Amazonian forest composition was altered by humans through the incipient domestication of tree species, increasing the abundance of a small number of favoured species42. Others have reported large areas of deforestation associated with the African Iron Age43. How can such different human histories result in near-identical patterns of tree species dominance? The most parsimonious explanation is that the system tends to return to a state with a similar species abundance pattern.
Nevertheless, consistent patterns of commonness do not necessarily imply the same causal mechanisms. The ubiquity of the broad ‘rare is common, common is rare’ pattern in ecology, which is also found in non-biological complex systems44, means inferences as to the cause of this broad pattern are challenging27,45. Although combinatoric methods45 and models that maximize the entropy of information46,47 both produce the ubiquitous ‘reverse lazy-J’ pattern, empirical observations show fewer common species and more rare species than expected by statistical controls alone45. Similarly, neutral models produce the same broad pattern, but produce too few individuals of the most common Amazonian tree species48. This suggests that biological mechanisms influence tree community assembly to produce a consistent proportion of common species across continents.
Recent analyses have revealed that the same few families contribute most of the species richness in Africa and Amazonia49, which when combined with analyses showing that more diverse families have more common species50, may indicate a role for deep evolutionary mechanisms driving the patterns we find. Yet, considering the substantially smaller regional species pool in Africa compared with Amazonia and Southeast Asia, one might expect differing continental patterns of species dominance if evolutionary drivers were the primary mechanism, not the highly consistent patterns that we find. Similarly, if environmental filtering were a key mechanism, the different contemporary environments, with Africa much drier on average than the other two continents26, and Southeast Asia consisting of scattered island-like areas of forest compared with the contiguous forested region of Amazonia, would also imply differing continental patterns of species dominance, not the near-identical patterns that we find. These constraints limit the potential mechanisms that could apply across our three-continent context.
One potential cross-continental mechanism is dispersal limitation, where the dispersal capabilities of species result in some suitable habitat patches remaining unoccupied. Another mechanism is density- or distance-dependent mortality, which appears widespread across tropical forests51. Here, specialist species-specific natural enemies such as pathogens and herbivores reduce seed or juvenile conspecific survival rates near conspecific adults or in areas of high juvenile conspecific density, thereby reducing competitive exclusion and contributing to the maintenance of high tree species richness in tropical forests51. It is possible that common species have largely evaded density- and/or distance-dependent mortality. Analyses showing that species abundance can be either high or low within given genera52 support this hypothesis. Further progress on putative mechanisms can be made, for example, by exploring whether ecological or functional traits differ between common and rare species, and assessing the consistency of any differences among tropical continents53. Although deducing mechanisms is complex, the identification of a tractable number of common species in tropical forests will facilitate progress in understanding of tropical forests beyond species abundance distributions.
Refining our results, particularly the naming of common species, requires improved sampling of tropical forests, both in terms of geographic scope and taxonomic identification of trees within plots. Expanding sampling to include Central America, New Guinea, Micronesia and other regions would improve the generality of our results. Better identifying trees in existing plots would increase the utility of available samples: in our Southeast Asia region we excluded 142 plots (approximately 120,000 stems) because they did not have more than 80% of trees identified to species. Furthermore, additional taxonomic research on even the most common species is needed given that some of the most common Amazonian33 and African54,55 tree species have been found to be complexes of several distinct species that are difficult to distinguish in the field. However, the similarity of our results across the three continental regions suggests that the occurrence of such species complexes may also be similar across the continental regions, again implying the operation of fundamental processes in differing forests. Overall, our work underscores the need for investment in taxonomy, particularly given the thousands of rare species we and others18 document, but also when considering the most common species.
Our best estimate, using extrapolation, that for the tropics as a whole just 1,053 species account for half of Earth’s 800 billion tropical trees has potentially profound implications. Rather than attempting to understand tens of thousands of species of tropical trees, a focus on just a few hundred of the most common species can provide a simplified characterization of these otherwise complex forests. Our analyses indicate that the most common of these species are reliably named and relatively well known. Our list of candidate hyperdominants can therefore readily serve new research, including in facilitating targeted autecological data collection to understand their role in providing ecological functions and services. Practically, this species-specific information could enhance tropical forest modelling by focusing on common species instead of relying on functional types or traits, thereby potentially improving predictions of future forest change.
In the future, analyses should be extended to investigate forest carbon stocks and hyperdominant species and their role in the provision of ecosystem services. In Amazonia, even fewer tree species were found to account for 50% of aboveground carbon stocks than the minimum number required to account for 50% of trees22. More generally, the set of common species is likely to include foundation species that define broader community assemblages, the environmental sensitivity of which will probably drive tropical forest responses to environmental change56. Of course, striving to understand and protect rare and non-hyperdominant species remains crucial, particularly as they face greater extinction risk and probably also contribute to the functioning of ecosystems, particularly when more functions57, longer timescales58 and imposed environmental changes59 are considered, and given that the hyperdominants of the future may be rarer today. Nonetheless, with a complementary grasp of the most common species, mapping, understanding and modelling of the world’s tropical forests will be a much more tractable proposition.
Methods
Data compilation and pre-processing
We collated data from forest inventory plots ≥0.2 ha in size, situated in structurally intact (no detectable past logging or fire), closed canopy (not dry forest or savanna) tropical forest, with enumeration of all stems ≥10 cm diameter, in which ≥ 80% of stems are identified to the species level. Following Sullivan et al.61, small (≤0.5 ha) plots within 1 km of each other were grouped for analysis to minimize the effect of stochastic tree fall events in smaller areas62. These criteria allow direct comparisons to be made with hyperdominance results from Amazonia6,21. The data from each continent comprise the following:
Africa: 483 plots, covering a total of 504 ha (mean plot area 1.04 ha, median 1 ha, range 0.2–10 ha). These data are from four sources: 299 plots from the African Tropical Rainforest Observatory Network1,63 (AfriTRON: www.afritron.org, accessed 1 March 2020), curated at http://www.ForestPlots.net64; 127 plots from the Central African Plot Network (https://central-african-plot-network.netlify.app); 52 plots from the TEAM network2; and 5 × 1 ha plots from 5 different soil types, extracted from one 50-ha plot in Korup, Cameroon from the SIGEO/CTFS network3.
Amazonia: 1,417 plots, covering a total of 1,591 ha (mean plot area 1.12 ha, median 1 ha, range 0.1–78.8 ha) from the Amazon Tree Diversity Network (ATDN: http://atdn.myspecies.info/, includes plots from the RAINFOR network), accessed 8 January 2020.
Southeast Asia: 230 plots, covering a total of 202 ha (mean plot area 0.88 ha, median 0.49 ha, range 0.21–4.5 ha). These data are from two sources: 143 plots from Slik et al.4,25—a decrease from the published Indo-Pacific dataset in Slik et al.4,25 due to our ≥80% species identification criterion and our Southeast Asia study region excluding Australia, India, and Papua New Guinea; and 87 plots from the T-Forces network64 curated at http://www.ForestPlots.net, accessed 03/02/2021.
Species names were checked for orthography and standardized (synonyms identified from the reference databases corrected to their accepted names) using the African Flowering Plants Database (https://www.ville-ge.ch/musinfo/bd/cjb/africa), Taxonomic Name Resolution Service65, and Asian Plant Synonym Lookup (F. Slik, personal communication), for Africa, Amazonia and Southeast Asia, respectively. Trees not identified to species level (7.3%, 6.3% and 8.4% of stems in the Africa, Southeast Asia datasets respectively) were classed as ‘indeterminate’ (Indet). Indet stems contributed to plot-level and dataset-wide stem abundance totals but are necessarily absent from species totals.
For the purposes of our study we delimited tropical forests according to the ‘tropical and subtropical moist broadleaf forests’ biome delineation from the World Wildlife Fund ecoregion map60. The total number of tropical trees ≥10 cm trunk diameter in each of our regions was then estimated by summing tree abundances in countries in which we have at least one sampled plot from the ‘map of Global Tree Density’66 (derived from 429,775 ground-based estimates of tree density) and masking according to the ‘tropical and subtropical moist broadleaf forests’ borders using ArcGIS v3.10.167. Thus, we estimate that there are ~92 billion, ~331 billion trees, and ~217 billion trees in our Africa, Amazonia, and Southeast Asia regions, respectively, totalling 640 billion trees. Including abundance from countries in the ‘tropical and subtropical moist broadleaf forests’ biome in which we have no sampled plots, we estimate ~799 billion total trees across all of Earth’s moist tropical forests.
Data format, commonness and diversity parameters
The species abundance distribution (SAD), defined as a vector of abundances (number of individuals observed) of all species encountered in a community17, formed the basis for our analyses of the three tropical forest datasets. For each dataset, we tallied the number of trees of each species in each plot to give plot-level SADs and combined these SADs across all plots to get regional-level abundance matrices with rows representing plots, columns representing species, and entries representing the abundance of each species in each plot. To capture patterns of commonness and species composition we calculated the number of hyperdominants (H#), defined as the minimum number of species required to account for 50% of the population of an assemblage6, hyperdominant species identities, total number of species (TS), hyperdominant percentage of total species (H% = H#/TS) and Fisher’s α (ref. 68). To investigate the sensitivity of results to the ‘hyperdominant’ definition of the most common species, we looked beyond the 50% threshold used for hyperdominance, at the minimum number of species required to account for 10%, 20%, 30%, …, 90% of the population, here termed ‘dominants’.
Sampling standardization, subsampling and comparison of continental data
We identified variations in the number of plots, stems, and species, and the size and spatial clustering of plots as potential confounding factors liable to skew dominance and diversity results from our regional datasets and impede rigorous comparisons between them. We used sample-based rarefaction to quantify and account for the effect of differences in sample size (number of plots and stems) on our diversity measures of interest; namely species richness, number, ranking and identity of hyperdominants, hyperdominant percentage of total species, and Fisher’s α. To quantify the effect of plot size, which is smaller in Southeast Asia data (mean 0.88 ha, median 0.49 ha) than in Amazonia and Africa data (both mean ~1 ha, median 1 ha) we compared results from the full data to those from plots >0.9 ha. We found that small plots (≪1 ha) inflate per-plot species totals relative to larger plots (because the rate of encountering new species is higher the smaller the plot size; Extended Data Fig. 1), so we limited our analyses to plots >0.9 ha to enable like-for-like comparison.
For Africa, we retained 368 plots covering 450 ha (mean plot area 1.22 ha, median 1 ha, range 0.92–10 ha; 2% of plots 0.9–0.99 ha, 88% of plots 1 ha, 8% of plots 1.01–5 ha, 1% of plots >5 ha) with mean temperature of 24.3 °C (range 16.2–27.6 °C), mean annual precipitation 1,802 mm yr−1, (range 1,066–2,747 mm yr−1), and mean elevation of 511 m above sea level (range 41–2,070 m) per WorldClim69. For Amazonia we retained 1,097 plots covering 1,434 ha (mean plot area 1.31 ha, median 1 ha, range 0.9–78.8 ha; 2% of plots 0.9–0.99 ha, 90% of plots 1 ha, 7% of plots 1.01–5 ha, 1% of plots >5 ha) with mean temperature of 26.0 °C (range 20.9–27.6 °C), mean annual precipitation 2,397 mm yr−1 (range 1,119–4,284 mm yr−1), and mean elevation of 154 m (range 0–1,142 m). For Southeast Asia we retained 103 plots covering 164 ha (mean plot area 1.59 ha, median 1 ha, range 0.96–4.5 ha; 1% of plots 0.9–0.99 ha, 48% of plots 1 ha, 52% of plots 1.01–5 ha, 0% of plots >5 ha) with mean temperature of 25.7 °C (range 20.1–27.5 °C), mean precipitation 2,680 mm yr−1 (range 1,466–3,941 mm yr−1), and mean elevation of 288 m (range 10–934 m). We assessed if the remaining differences in plot size affected the results, using only the 1 ha plots from Africa (n = 323) and Amazonia (n = 988), rarefied to the size of the Asia dataset, again finding near-identical per cent hyperdominance on the two continents (Africa: 7.30%, 95% confidence interval: 6.56–8.04; Amazonia: 7.35%, 95% confidence interval: 6.61–8.10).
To quantify the effect of the spatial clustering of plots, we compared results from the full Amazonia data, as the largest dataset, to those from subsets of the Amazonia data in which 1,2,3,…,10 plots were sampled from each spatial cluster. We found that spatial clustering had a negligible and not statistically significant effect on hyperdominant percentage and fitted values of Fisher’s α (Extended Data Fig. 2). Therefore, we retain all plots for our analyses to maximize sample sizes. Computation of percentage hyperdominance and dominance accounts for the effects of variations in species richness on the number of hyperdominants and dominants.
For sample-based rarefaction, 200 subsamples of 1, 2, …, Np plots were drawn, without replacement, from the Np total number of plots in the pth dataset, the stems contained in each subsample were pooled, and the mean total species, number of hyperdominants, hyperdominance percentage, and Fisher’s α were calculated across the subsamples. Similarly, we tallied the number of subsamples in which each species in the dataset qualified as hyperdominant at each level of subsampling and compared results between datasets at subsample sizes equating to a mean 10,000, 20,000, …, Ip individual trees, where Ip is the total number of trees in the pth dataset. Confidence intervals were calculated as confidence interval = μ ± 1.96 × σ, where μ values are the means of the diversity metrics calculated across the 200 iterations of subsamples taken without replacement, and σ values are the s.d. of the mean of diversity metrics calculated across the 200 iterations of subsamples taken with replacement (to reduce the degree to which confidence intervals were conditional on the sample). For point estimates, all datasets were compared at the common sample size of the Southeast Asia dataset (77,587 stems equivalent to 150, 116 and 103 plots in Africa, Amazonia and Southeast Asia, respectively).
Extrapolation and bias correction of log series fits to the empirical data
We extrapolated our empirical SADs to SADs at the scale of the entire Amazonian, African, and Southeast Asian regional level via analytical expansion and bias correction of Fisher’s log series fits following the methodology of ter Steege et al.21 developed using the ATDN data that comprise our Amazonia dataset.
Ter Steege21 et al. found that simulations of sampling of plots with conspecific aggregation from log series-modelled SADs provide extremely good approximations of the processes that generate tropical forest inventory data—that is, non-random sampling of plots containing species with limited dispersal and/or ecological preferences. They further found that estimates of species richness derived from samples taken with conspecific aggregation from the simulated SADs substantially underestimated the true species richness of the simulated SADs, but that a linear relationship with low variance existed between the true and sample-derived values. Thus, although conspecific aggregation in the empirical data introduces bias in the log series-modelled SADs extrapolated therefrom, quantification and correction of the effects of this bias on regional estimates of species richness is possible. Therefore, to estimate species richness at the regional level, they fitted Fisher’s log series to empirical species abundance data, quantified the effect of conspecific aggregation on these estimates via simulation, and applied quantified corrections to give more accurate estimates of regional species richness taking into conspecific aggregation. Thus, this approach corrects for species-specific aggregation at the plot scale depending on species density.
To estimate regional numbers and proportions of dominants and hyperdominants as well as species richness, we extended the methodology of ter Steege et al.21 to log series-derived estimates of regional numbers and proportions of dominants and hyperdominants. Initially, values of Fisher’s α were fitted to the empirical species abundance vectors from each region using maximum likelihood and numerical optimization in the ‘sads’ R package70 and fits visualized with Preston plots71 and rank abundance distributions (RAD)36 (Extended Data Fig. 4). Regional species totals S, not accounting for bias introduced by conspecific aggregation, were then estimated68 via \(S=\alpha \times {\rm{ln}}\left(1+\frac{N}{\alpha }\right)\) with total number of trees ≥10 cm trunk diameter at the continental level (N) from the Global Tree Density map of Crowther et al.66 with each tropical region delineated within the ‘tropical and subtropical moist broadleaf forests’ biome of Olson et al.60. An inverse quantile function from the sads R package70 was then applied to generate (uncorrected) continental-scale SADs for each region using the above fitted α, estimated S and N.
For the quantification of bias and computation of corrections, we first simulated 250 log series SADs with known values of total species, Sk, randomly drawn from the range of plausible regional species totals (10,000–25,000 in Amazonia and Southeast Asia; 2,000–10,000 in Africa) and N, the number of trees in each region ≥10 cm trunk diameter from Crowther et al.66. We calculated known values of numbers of hyperdominants, Hk, and percentage hyperdominance, Pk, from each of these simulated distributions. Using a negative binomial distribution to simulate conspecific aggregation per ter Steege et al.21, we then simulated j random samples of 1-ha plots from each of the 250 simulated SADs, with j equal to the number of plots in the empirical data, and the expected abundance of each species in each plot equal to its mean regional density (total abundance/regional area). We then estimated (uncorrected) species richness, Su, from each of the samples by fitting Fisher’s α to the sampled data and applying the formula \({S}_{u}=\alpha \times {\rm{ln}}\left(1+\frac{N}{\alpha }\right)\). From each of the samples we also derived continental-scale uncorrected SADs (see above), from which the number of hyperdominants, Hu, and percentage hyperdominance, Pu, could be directly calculated, via analytical expansion of the log series using the fitted values of α and corresponding values of Su. We then regressed the known values of Sk, Hk and Pk from the simulated SADs against the estimated (uncorrected) values Su, Hu and Pu from the samples drawn with conspecific aggregation across all 250 simulations—that is, fit linear models of the form Ak = m × Au + c for A = S, H, P. This same procedure was also applied to the number and proportion of dominants.
Across all three regional datasets, the above procedure outlined a linear relationship with low variance between known values of species richness, number of dominants and hyperdominants, and percentage hyperdominance and dominance, and values thereof estimated from sampling with conspecific aggregation (Extended Data Fig. 5). Thus, constant terms with low variance were readily applicable to correct for bias in the point estimates of species richness, number of dominants/hyperdominants, and percentage hyperdominance/dominance, derived from the empirical Africa, Amazonia, and Southeast Asia data. To capture uncertainty around each bias-corrected point estimate, prediction intervals (PI) were derived as PI = μ + 1.96 × σPI, where μ is the predicted mean value of the point estimate according to the linear regression, and σPI is the PI standard error, calculated as \({\sigma }_{{\rm{PI}}}=\sqrt{{\sigma }^{2}+{\sigma }_{{\rm{R}}}^{2}}\), where σ is the standard error of predicted means and σR is the residual s.d. (and 1.96 is the 0.05 quantile of a t-distribution).
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The species abundance data that support the findings of this study are available from https://doi.org/10.6084/m9.figshare.21670883 (formatting notes: a column for each species, rows for each plot, entries are the number of trees ≥10 cm diameter of each species in each plot). WorldClim69 bioclimatic data are available from https://www.worldclim.org/data/bioclim.html.
Code availability
R code (version 4.3.1) to run the analyses and produce the figures and tables is available from https://github.com/declancooper/CommonSpecies2022.git.
References
Lewis, S. L. et al. Above-ground biomass and structure of 260 African tropical forests. Phil. Trans. R. Soc. B. 368, 20120295 (2013).
Rovero, F. & Ahumada, J. The Tropical Ecology, Assessment and Monitoring (TEAM) Network: an early warning system for tropical rain forests. Sci. Total Environ. 574, 914–923 (2017).
Anderson‐Teixeira, K. J. et al. CTFS–Forest GEO: a worldwide network monitoring forests in an era of global change. Glob. Change Biol. 21, 528–549 (2015).
Slik, J. W. F. et al. Phylogenetic classification of the world’s tropical forests. Proc. Natl Acad. Sci. USA 115, 1837–1842 (2018).
Qie, L. et al. Long-term carbon sink in Borneo’s forests halted by drought and vulnerable to edge effects. Nat. Commun. 8, 1966 (2017).
ter Steege, H. et al. Hyperdominance in the Amazonian tree flora. Science 342, 1243092 (2013).
Corlett, R. T. & Primack, R. B. Tropical Rain Forests: An Ecological and Biogeographical Comparison (John Wiley & Sons, 2011).
IPCC Climate change 2022. Impacts, Adaptation and Vulnerability (eds Pörtner, H.-O. et al.) (Cambridge Univ. Press., 2022).
Gough, C. Terrestrial primary production: fuel for life. Nat. Educ. Knowl. 3, 28 (2011).
Erb, K.-H. et al. Unexpectedly large impact of forest management and grazing on global vegetation biomass. Nature 553, 73–76 (2018).
Hubau, W. et al. Asynchronous carbon sink saturation in African and Amazonian tropical forests. Nature 579, 80–87 (2020).
Dirzo, R. & Raven, P. H. Global state of biodiversity and loss. Annu. Rev. Env. Res. 28, 137–167 (2003).
Mittermeier, R. A., Turner, W. R., Larsen, F. W., Brooks, T. M. & Gascon, C. in Biodiversity Hotspots (eds Zachos, F. & Habel, J.) 3–22 (Springer, 2011).
Valencia, R., Balslev, H. & Paz Y Miño C, G. High tree alpha-diversity in Amazonian Ecuador. Biodivers. Conserv. 3, 21–28 (1994).
Beech, E., Rivers, M., Oldfield, S. & Smith, P. P. GlobalTreeSearch: the first complete global database of tree species and country distributions. J. Sustain. For. 36, 454–489 (2017).
Cazzolla Gatti, R. et al. The number of tree species on Earth. Proc. Natl Acad. Sci. USA 119, e2115329119 (2022).
McGill, B. J. et al. Species abundance distributions: moving beyond single prediction theories to integration within an ecological framework. Ecol. Lett. 10, 995–1015 (2007).
Enquist, B. J. et al. The commonness of rarity: global and future distribution of rarity across land plants. Sci. Adv. 5, eaaz0414 (2019).
Baldridge, E., Harris, D. J., Xiao, X. & White, E. P. An extensive comparison of species-abundance distribution models. PeerJ 4, e2823 (2016).
Draper, F. C. et al. Amazon tree dominance across forest strata. Nat. Ecol. Evol. 5, 757–767 (2021).
ter Steege, H. et al. Biased-corrected richness estimates for the Amazonian tree flora. Sci. Rep. 10, 10130 (2020).
Fauset, S. et al. Hyperdominance in Amazonian forest carbon cycling. Nat. Commun. 6, 6857 (2015).
Pitman, N. C. A., Silman, M. R. & Terborgh, J. W. Oligarchies in Amazonian tree communities: a ten-year review. Ecography 36, 114–123 (2013).
Pitman, N. C. A. et al. Dominance and distribution of tree species in upper Amazonian terra firme forests. Ecology 82, 2101–2117 (2001).
Slik, J. W. et al. An estimate of the number of tropical tree species. Proc. Natl Acad. Sci. USA 112, 7472–7477 (2015).
Parmentier, I. et al. The odd man out? Might climate explain the lower tree α‐diversity of African rain forests relative to Amazonian rain forests? J. Ecol. 95, 1058–1071 (2007).
McGill, B. J. & Nekola, J. C. Mechanisms in macroecology: AWOL or purloined letter? Towards a pragmatic view of mechanism. Oikos 119, 591–603 (2010).
Richards, P. W. in Tropical Forest Ecosystems of Africa and South America: A Comparative Review (eds Meggers, B. J., Ayensu, E. S. & Duckworth, W. D.) 21–26 (Smithsonian Institution Press, 1973).
Couvreur, T. L. Odd man out: why are there fewer plant species in African rain forests? Plant Syst. Evol. 301, 1299–1313 (2015).
Tovo, A. et al. Upscaling species richness and abundances in tropical forests. Sci. Adv. 3, e1701438 (2017).
Cardoso, D. et al. Amazon plant diversity revealed by a taxonomically verified species list. Proc. Natl Acad. Sci. USA 114, 10695–10700 (2017).
Ter Steege, H. et al. Towards a dynamic list of Amazonian tree species. Sci. Rep. 9, 3501 (2019).
Damasco, G. et al. Revisiting the hyperdominance of Neotropical tree species under a taxonomic, functional and evolutionary perspective. Sci. Rep. 11, 9585 (2021).
Kattge, J. et al. TRY plant trait database—enhanced coverage and open access. Glob. Change Biol. 26, 119–188 (2020).
Darwin, C. On The Origin of Species by Means of Natural Selection: Or, the Preservation of Favored Races in the Struggle for Life (J. Murray, 1859).
McGill, B. J. in Biological Diversity: Frontiers In Measurement and Assessment (ed. Magurran, A. E. & McGill, B. J.) 105–122 (2011).
Henderson, P. A. & Magurran, A. E. Linking species abundance distributions in numerical abundance and biomass through simple assumptions about community structure. Proc. R. Soc. B 277, 1561–1570 (2010).
Currano, E., Jacobs, B. & Pan, A. Is Africa really an “odd man out”? Evidence for diversity decline across the Oligocene–Miocene boundary. Int. J. Plant Sci. 182, 551–563 (2021).
Morley, R. J. Origin and Evolution of Tropical Rain Forests (John Wiley & Sons, 2000).
Scerri, E. M. L., Roberts, P., Maezumi, S. Y. & Malhi, Y. Tropical forests in the deep human past. Phil. Trans. R. Soc. B 377, 20200500 (2022).
Sandom, C., Faurby, S., Sandel, B. & Svenning, J.-C. Global late Quaternary megafauna extinctions linked to humans, not climate change. Proc. R. Soc. B 281, 20133254 (2014).
Levis, C. et al. Persistent effects of pre-Columbian plant domestication on Amazonian forest composition. Science 355, 925–931 (2017).
Garcin, Y. et al. Early anthropogenic impact on Western Central African rainforests 2,600 y ago. Proc. Natl Acad. Sci. USA 115, 3261–3266 (2018).
Nekola, J. C. & Brown, J. H. The wealth of species: ecological communities, complex systems and the legacy of Frank Preston. Ecol. Lett. 10, 188–196 (2007).
Diaz, R. M., Ye, H. & Ernest, S. K. M. Empirical abundance distributions are more uneven than expected given their statistical baseline. Ecol. Lett. 24, 1739–2039 (2021).
Harte, J. & Newman, E. A. Maximum information entropy: a foundation for ecological theory. Trends Ecol. Evol. 29, 384–389 (2014).
Harte, J., Brush, M., Newman, E. A. & Umemura, K. An equation of state unifies diversity, productivity, abundance and biomass. Commun. Biol. 5, 874 (2022).
Pos, E. et al. Scaling issues of neutral theory reveal violations of ecological equivalence for dominant Amazonian tree species. Ecol. Lett. 22, 1072–1082 (2019).
Silva de Miranda, P. L. et al. Dissecting the difference in tree species richness between Africa and South America. Proc. Natl Acad. Sci. USA 119, e2112336119 (2022).
Webb, C. O. & Pitman, N. C. Phylogenetic balance and ecological evenness. Syst. Biol. 51, 898–907 (2002).
Comita, L. S. et al. Testing predictions of the Janzen–Connell hypothesis: a meta‐analysis of experimental evidence for distance‐and density‐dependent seed and seedling survival. J. Ecol. 102, 845–856 (2014).
Ricklefs, R. E. & Renner, S. S. Global correlations in tropical tree species richness and abundance reject neutrality. Science 335, 464–467 (2012).
Koffel, T., Umemura, K., Litchman, E. & Klausmeier, C. A. A general framework for species‐abundance distributions: linking traits and dispersal to explain commonness and rarity. Ecol. Lett. 25, 2359–2371 (2022).
Ikabanga, D. U. et al. Combining morphology and population genetic analysis uncover species delimitation in the widespread African tree genus Santiria (Burseraceae). Phytotaxa 321, 166 (2017).
Koffi, K. G. et al. A combined analysis of morphological traits, chloroplast and nuclear DNA sequences within Santiria trimera (Burseraceae) suggests several species following the Biological Species Concept. Plant Ecol. Evol. 143, 160–169 (2010).
Ellison, A. M. et al. Loss of foundation species: consequences for the structure and dynamics of forested ecosystems. Front. Ecol. Environ. 3, 479–486 (2005).
Lefcheck, J. S. et al. Biodiversity enhances ecosystem multifunctionality across trophic levels and habitats. Nat. Commun. 6, 6936 (2015).
Isbell, F. et al. Quantifying effects of biodiversity on ecosystem functioning across times and places. Ecol. Lett. 21, 763–778 (2018).
Isbell, F. et al. High plant diversity is needed to maintain ecosystem services. Nature 477, 199–202 (2011).
Olson, D. M. et al. Terrestrial Ecoregions of the World: a new map of life on Earth: a new global map of terrestrial ecoregions provides an innovative tool for conserving biodiversity. Bioscience 51, 933–938 (2001).
Sullivan, M. J. P. et al. Long-term thermal sensitivity of Earth’s tropical forests. Science 368, 869–874 (2020).
Clark, D. B. & Clark, D. A. Landscape-scale variation in forest structure and biomass in a tropical rain forest. For. Ecol. Manag. 137, 185–198 (2000).
ForestPlots.net.Taking the pulse of Earth’s tropical forests using networks of highly distributed plots. Biol. Conserv. 260, 108849 (2021).
Lopez‐Gonzalez, G., Lewis, S. L., Burkitt, M. & Phillips, O. L. ForestPlots.net: a web application and research tool to manage and analyse tropical forest plot data. J. Veg. Sci. 22, 610–613 (2011).
Boyle, B. et al. The taxonomic name resolution service: an online tool for automated standardization of plant names. BMC Bioinformatics 14, 16 (2013).
Crowther, T. W. et al. Mapping tree density at a global scale. Nature 525, 201–205 (2015).
Esri, A. D. ArcGIS Release 10. Documentation Manual (Environmental Systems Research Institute, 2011).
Fisher, R. A., Corbet, A. S. & Williams, C. B. The relation between the number of species and the number of individuals in a random sample of an animal population. J. Anim. Ecol. 12, 42–58 (1943).
Fick, S. E. & Hijmans, R. J. WorldClim 2: new 1‐km spatial resolution climate surfaces for global land areas. Int. J. Climatol. 37, 4302–4315 (2017).
Prado, P. I., Miranda, M. D., Chalom, A., Prado, M. P. I. & Imports, M. sads: maximum likelihood models for species abundance distributions. R package version 0.4.2 (2018).
Preston, F. W. The commonness, and rarity, of species. Ecology 29, 254–283 (1948).
Acknowledgements
D.L.M.C. was supported by the London Natural Environmental Research Council Doctoral Training Partnership grant (grant no. NE/L002485/1). This paper developed from analysing data from the African Tropical Rainforest Observatory Network (AfriTRON), curated at ForestPlots.net. AfriTRON has been supported by numerous people and grants since its inception. We sincerely thank the people of the many villages and local communities who welcomed our field teams and without whose support this work would not have been possible. Grants that have funded the AfriTRON network, including data in this paper, are a European Research Council Advanced Grant (T-FORCES; 291585; Tropical Forests in the Changing Earth System), a NERC standard grant (NER/A/S/2000/01002), a Royal Society University Research Fellowship to S.L.L., a NERC New Investigators Grant to S.L.L., a Philip Leverhulme Award to S.L.L., a European Union FP7 grant (GEOCARBON; 283080), Leverhulme Program grant (Valuing the Arc); a NERC Consortium Grant (TROBIT; NE/D005590/), NERC Large Grant (CongoPeat; NE/R016860/1) the Gordon and Betty Moore Foundation the David and Lucile Packard Foundation, the Centre for International Forestry Research (CIFOR), and Gabon’s National Parks Agency (ANPN). This paper was supported by ForestPlots.net approved Research Project 81, ‘Comparative Ecology of African Tropical Forests’. The development of ForestPlots.net and data curation has been funded by several grants, including NE/B503384/1, NE/N012542/1, ERC Advanced Grant 291585—‘T-FORCES’, NE/F005806/1, NERC New Investigators Awards, the Gordon and Betty Moore Foundation, a Royal Society University Research Fellowship and a Leverhulme Trust Research Fellowship. Fieldwork in the Democratic Republic of the Congo (Yangambi and Yoko sites) was funded by the Belgian Science Policy Office BELSPO (SD/AR/01A/COBIMFO, BR/132/A1/AFRIFORD, BR/143/A3/HERBAXYLAREDD, FED-tWIN2019-prf-075/CongoFORCE, EF/211/TREE4FLUX); by the Flemish Interuniversity Council VLIR-UOS (CD2018TEA459A103, FORMONCO II); by L’Académie de recherche et d’enseignement supérieur ARES (AFORCO project) and by the European Union through the FORETS project (Formation, Recherche, Environnement dans la TShopo) supported by the XIth European Development Fund. EMV was supported by fellowship from the CNPq (Grant 308543/2021-1). RAPELD plots in Brazil were supported by the Program for Biodiversity Research (PPBio) and the National Institute for Amazonian Biodiversity (INCT-CENBAM). BGL post-doc grant no. 2019/03379-4, São Paulo Research Foundation (FAPESP). D.A.C. was supported by the CCI Collaborative fund. Plots in Mato Grosso, Brazil, were supported by the National Council for Scientific and Technological Development (CNPq), PELD-TRAN 441244/2016-5 and 441572/2020-0, and Mato Grosso State Research Support Foundation (FAPEMAT)—0346321/2021. We thank E. Chezeaux, R. Condit, W. J. Eggeling, R. M. Ewers, O. J. Hardy, P. Jeanmart, K. L. Khoon, J. L. Lloyd, A. Marjokorpi, W. Marthy, H. Ntahobavuka, D. Paget, J. T. A. Proctor, R. P. Salomão, P. Saner, S. Tan, C. O. Webb, H. Woell and N. Zweifel for contributing forest inventory data. We thank numerous field assistants for their invaluable contributions to the collection of forest inventory data, including A. Nkwasibwe, ITFC field assistant.
Author information
Authors and Affiliations
Contributions
D.L.M.C. and S.L.L. conceived and developed the study. D.L.M.C. performed the analysis with M.J.P.S. and P.I.P. and input from S.L.L. D.L.M.C., P.I.P., G.C.P., A.L. and M.J.P.S. developed tools to support the analysis. D.L.M.C. and S.L.L. wrote the manuscript with significant input from M.J.P.S., R.G.P. and M.I.D. S.L.L., B.S. and C.E.N.E. curated the AfriTRON forest plot data. N.B., P.P. and G.D. curated the Central African Plot Network forest plot data. H.T.S. curated the ATDN forest plot data. F.S. curated the Slik et al.4,25 Southeast Asia forest plot data. S.L.L. and O.L.P. curated the T-FORCES Southeast Asia carbon monitoring network. O.L.P., S.L.L., T.R.B., B.S, B.S.M., C.E., E.H., L.Q., A.L. and G.P. provided ForestPlots.net pan-tropical data management. All co-authors contributed data, reviewed, approved and had the opportunity to comment on the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature thanks Miguel Martinez-Ramos, Thorsten Wiegand and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer review reports are available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Extended Data Fig. 1 Impact of plot size on rarefaction curves of total species (a) and number of hyperdominants (b) in the Asia data.
Red points represent the full Southeast Asia data (mean values across iterations of subsamples), including all plot sizes (mean plot size: 0.877 ha, median plot size: 0.5 ha); Purple points represent the Southeast Asia data restricted to plots ≥0.9 ha (mean plot size: 1.59 ha, median plot size: 1 ha).
Extended Data Fig. 2 Impact of spatial clustering of plots on rarefaction curves of hyperdominant percentage (first row) and Fisher’s Alpha (second row) in the Amazonia data.
Purple points and confidence intervals represent the full data; black points and confidence intervals represent a subset of the data in which one plot is sampled from each spatial cluster of plots; other coloured points represent subsets of the data in which 2,3,4,…,10 plots (or the total number of plots in the cluster) are sampled from each spatial cluster of plots. Points give the mean values across iterations of subsamples. Confidence intervals are derived via the standard deviation across iterations of subsamples taken with replacement at each sampling point. Note that although resampling for rarefaction was done by subsampling tree inventory plots, the curves are re-plotted with an x-axis of number of stems.
Extended Data Fig. 3 Complete rarefaction curves showing the effect of increasing sampling on the number of hyperdominants (a), total species (b), hyperdominant percentage (c), and fitted values of Fisher’s α (d).
In tropical Africa (magenta), Amazonia (cyan), Southeast Asia (blue). Markers represent rarefied points (mean values across iterations of subsamples); shaded areas represent confidence intervals (CIs). Confidence intervals are derived via the standard deviation across iterations of subsamples taken with replacement at each sampling point. Note that although resampling for rarefaction was done by subsampling tree inventory plots, the curves are re-plotted with an x-axis of number of stems.
Extended Data Fig. 4 Preston plots (top row) and rank abundance distributions (bottom row) showing the empirical species abundance distributions for Africa (left) Amazonia (middle) and Southeast Asia (right) with log series fits overlaid.
Histogram bars display the empirical species abundance distributions as Preston plots (top row); black markers show the empirical species abundance distributions as rank abundance distributions (bottom row); overlaid points and lines show log series fits to empirical species abundance distributions in Africa (magenta), Amazonia (cyan), and Southeast Asia (blue).
Extended Data Fig. 5 Bias correction of estimates of species richness (first column), number of hyperdominants (second column), percentage hyperdominance (third column) for the Amazonia (first row), Africa (second row) and Southeast Asia (third row) datasets.
X-axes show estimated values derived from samples of the simulated communities taken with conspecific aggregation, Y-axes show true values of the simulated communities. Points show estimated true values for each of the 250 simulated communities. 1:1 equivalence shown by straight line in each plot. For number of hyperdominants and total species plots, simulated communities containing 100 to 25,000 species in Amazonia and Southeast Asia, 100 to 10,000 species in Africa are shown. For percentage hyperdominance, simulated communities containing 10,000 to 25,000 species in Amazonia and Southeast Asia, 2,000 to 10,000 species in Africa are shown.
Supplementary information
Supplementary Table 1
Percentage of subsamples in which each species in the Amazonia, Africa, and Southeast Asia empirical data qualifies as hyperdominant. Rows represent species, ranked by abundance in the empirical sample; columns represent number of stems in subsample; entries represent the percentage of the iterations of each level of subsampling in which each species qualified as hyperdominant, color-coded from green (high hyperdominant proportion of subsamples) to yellow (intermediate hyperdominant proportion of subsamples) to red (low hyperdominant proportion of subsamples). Lines delimit four groups: 1 hyperdominant in the full data and the majority of subsamples, even at very small sample sizes, 2 hyperdominant in the full data and the majority of subsamples, but only occasionally hyperdominant at small sample sizes, 3 not hyperdominant in the full data but hyperdominant in a substantial proportion of subsamples, and 4 not hyperdominant in the full data and almost never hyperdominant in subsamples.
Supplementary Table 2
List of 1,119 candidate hyperdominant tree species across the pan-tropics. Rows represent species, ranked by abundance in the empirical sample.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Cooper, D.L.M., Lewis, S.L., Sullivan, M.J.P. et al. Consistent patterns of common species across tropical tree communities. Nature 625, 728–734 (2024). https://doi.org/10.1038/s41586-023-06820-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41586-023-06820-z
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
