Posts Tagged ‘biomes’

We need mechanistic vegetation models!

August 19th, 2023 No comments

The journal Science recently rejected reply letters from at least 3 groups of top scientists to a poor paper they published (Higgins et al. 2023). Science only accepted comments as footnotes to the original article (termed eLetters). It is disappointing that a top journal like Science published a modelling paper that does not simulate the most important mechanisms of the system the authors aim to model. However, it is even more disappointing that Science is not more open to scientific debate, which is an important part of the scientific process. Therefore, we have added a brief comment to the poor paper as an eLetter in Science (also available in PDF) as the only way to warn the readers of some shortcomings (there are more that we hope will be published elsewhere). Below is a copy of our eLetter. 

We need mechanistic models to explain Alternative Ecosystem States in tropical vegetation

by William J. Bond & Juli G. Pausas

In their paper Limited climatic space for alternative ecosystem states in Africa (Science 8 June 2023, p. 1038), Higgins et al. use a plant growth model applied to species distribution and climate variables, to argue that Alternative Ecosystem States (AES) have limited importance in Africa. However, their model does not account for key ecological factors in Africa such as large herbivores and fires (1, 2). Their exclusion raises serious doubts about the model’s validity. Higgins et al. emphasise how well their model predicts the distribution of forests and savannas. However there is a poor fit in the maps predicting optimal areas for several growth forms (3, 4). The model failed to identify the major areas of shrub dominance in Africa, fynbos and karoo shrublands in the south-west and steppe shrublands in north-east Africa (their Figure 1). Succulents are predicted for the north African Mediterranean coast where there are none (and for large areas of southern Africa) presumably by erroneous extrapolation from succulent distribution in South Africa. Optimal ‘relative climatic suitability’ for C4 grasses is predicted for the southern margins of the Sahara desert (their Figure S2) but not the vast savannas that cover most of the rest of the continent.

Higgins et al. argue that the maps they derived show limited potential for tree growth in areas they identified as climatically limited savannas contradicting other studies identifying large areas of mesic savannas as suitable for large-scale tree planting. However, their model fails to predict a forest-suitable environment in areas supporting large-scale commercial forestry plantations, e.g. in southern Africa (5). In fact, it is unknown how much of the areas they predict as savannas, and are actually savannas (“true savanna predictions”; in their Fig. 3), could sustain a forest, and would therefore be examples of AES. Higgins et al. gloss over additional evidence for AES including paleoecological studies of system shifts between savanna and forest, hysteresis, historical studies, remote sensing and multi-decadal fire suppression experiments, both natural and by design, in Africa and elsewhere showing major ecosystem shifts typically linked to fire suppression or addition (6-10).

We conclude that Higgins et al. cannot be used as a basis for interpreting alternative ecosystem states, the potential for tree planting in Africa, or whether climate and physical site factors determine forest and savanna distribution. We suggest that the problem may lie in assuming that the distribution of species represents a fundamental niche and not a realised niche so that their apparently physiologically based model is really a rather complex correlative model following a long line of predecessors. The models lack seedling and sapling stages, widely considered to be key to whether trees can escape the flame zone and thereby exist in savannas or be restricted to forests (11). Fire is not included as a source of biomass loss. Nor is there any explicit consideration of shade, a major factor separating forest from savanna species (12). It lacks most of the fundamental mechanism to simulate a dynamic system such as African ecosystems. Exploration of the dynamic response of the model, for example to changing CO2 from 400 ppm to 280, might help reveal its sensitivity to environmental drivers outside those used to derive physiological parameters from inverse models of contemporary plant species distributions. Process-based models based on measured physiological, and fire response traits, are more appropriate tools for exploring the potential for alternative stable states because they test what could be and are not restricted by what is (13-15).

Higgins et al. model contributes little to the understanding of the processes assembling African ecosystems, and cannot be taken as evidence against AES. In our changing world, we need more mechanistic and dynamic models before casting aside all evidence for fire and herbivores limiting distributions of forests (2).


  1. N. Owen-Smith, Only in Africa: The Ecology of Human Evolution (Cambridge University Press, Cambridge, 2021).
  2. W. J. Bond, Open Ecosystems: Ecology and Evolution Beyond the Forest Edge (Oxford University Press, 2019).
  3. F. White, “The vegetation of Africa: a descriptive memoir to accompany the UNESCO/AETFAT/UNSO vegetation map of Africa by F White” (Natural Resources Research Report XX, UNESCO, Paris, France, 1983), pp. 1876-1895.
  4. D. A. Keith, J. R. Ferrer-Paris, E. Nicholson, R. T. Kingsford, Eds., IUCN Global Ecosystem Typology 2.0: descriptive profiles for biomes and ecosystem functional groups (IUCN, International Union for Conservation of Nature, 2020).
  5. Z. Du, L. Yu, J. Yang, Y. Xu, B. Chen, S. Peng, T. Zhang, H. Fu, N. Harris, P. Gong, A global map of planting years of plantations. Sci Data. 9, 141 (2022).
  6. L. Gillson, Evidence of a tipping point in a southern African savanna? Ecol. Complex. 21, 78-86 (2015).
  7. Z. S. Venter, M. D. Cramer, H. J. Hawkins, Drivers of woody plant encroachment over Africa. Nat. Commun. 9, 2272 (2018).
  8. J. C. Aleman, O. Blarquez, H. Elenga, J. Paillard, V. Kimpuni, G. Itoua, G. Issele, Staver A. Carla, Palaeo-trajectories of forest savannization in the southern Congo. Biol. Lett. 15, 20190284 (2019).
  9. J. G. Pausas, W. J. Bond, Alternative biome states in terrestrial ecosystems. Trends Plant Sci. 25, 250-263 (2020). 
  10. H. Beckett, A. C. Staver, T. Charles-Dominique, W. J. Bond, Pathways of savannization in a mesic African savanna-forest mosaic following an extreme fire. J. Ecol. 110, 902-915 (2022).
  11. C. P. Osborne, T. Charles-Dominique, N. Stevens, W. J. Bond, G. Midgley, C. E. R. Lehmann, Human impacts in African savannas are mediated by plant functional traits. New Phytol. 220, 10-24 (2018).
  12. T. Charles-Dominique, G. F. Midgley, K. W. Tomlinson, W. J. Bond, Steal the light: shade vs fire adapted vegetation in forest-savanna mosaics. New Phytol. 218, 1419-1429 (2018).
  13. W. J. Bond, F. I. Woodward, G. F. Midgley, The global distribution of ecosystems in a world without fire. New Phytol. 165, 525-538 (2005).
  14. S. I. Higgins, S. Scheiter, Atmospheric CO2 forces abrupt vegetation shifts locally, but not globally. Nature. 488, 209-212 (2012).
  15. G. Lasslop, S. Hantson, S. P. Harrison, D. Bachelet, C. Burton, M. Forkel, M. Forrest, F. Li, J. R. Melton, C. Yue, S. Archibald, S. Scheiter, A. Arneth, T. Hickler, S. Sitch, Global ecosystems and fire: multi-model assessment of fire-induced tree cover and carbon storage reduction. Global Change Biol. 26, 5027-5041 (2020).

“I would rather discover one cause than gain the kingdom of Persia”, Democritus (460-370 BC)

Update: Aleman and collaborators have also published an eLetter in response to the paper by Higgins. Here is their eLetter: link | PDF

Herbivores and fences

May 16th, 2023 No comments

Fenced areas are often illustrative of the role of herbivores in shaping vegetation [1]. Here is an example from Schönbuch Nature Park (south west of Stuttgart, Germany), where you can see how the vegetation is shaped by different densities of red deer (Cervus elaphus). Is this park-like landscape (Fig. 2; inside the enclosure) an example of the prehistoric lowland landscapes in Central Europe?

Fig. 1. Fence separating the area with absence (left) / presence (right) of red deer
Fig. 2. Within the enclosure where the density of red deer is highest. Is this park-like landscape an example of the prehistoric lowland landscapes in Central Europe?
Fig. 3. Hervivores of central Europe, from [2].

[1] Pausas JG & Bond WJ. 2020. Alternative biome states in terrestrial ecosystems. Trends Pl Sci, 25(3), 250–263. [doi | pdf

[2] Vera FWM. 2000. Grazing ecology and forest history. CAB International.

Alternative Biome States in Brazil

October 31st, 2022 No comments

I have recently been in Minas Gerais, Brazil (in the Cerrado region). I visited different biomes (forests, savannas, grasslands) occurring in the same climate, i.e., Alternative Biome States [1,2]. The sharp boundaries that separate the different biomes (photos blow) suggesting the existence of strong feedbacks [3]. Savannas and grasslands are maintained by frequent fires (flammable or pyrophilic communities) in climates where dense forest can occur; frequent fires maintain those open ecosystems dominated by light-demanding grasses, and woody plants with traits for fire survival (thick corky bark [4], epicormic resprouting [5], belowground organs [6]). In contrast, forest rarely get burn (non-flammable or pyrofobic communities), as the low light inhibit grasses and generate microclimatic conditions that are not favorable for fire (no grasses, high humidity, low wind, etc.) but favorable for shade-tolerant forest trees.

Forest-grassland mosaic in Serra da Canastra, Minas Gerais, Brazil
Forest-grassland mosaic in Serra da Canastra, Minas Gerais, Brazil
Savanna dominated by Vochysia thyrsoidea (Vochysiaceae; the large tree), in Serra da Canastra, Minas Gerais, Brazil. Brazilian savannas are often termed “cerrado”.


[1] Pausas JG & Bond WJ. 2020. Alternative biome states in terrestrial ecosystems. Trends Pl Sci 25: 250-263. [doi | sciencedirect | cell | pdf]

[2] Dantas VL, Hirota M, Oliveira RS, Pausas JG. 2016. Disturbance maintains alternative biome states. Ecol Lett 19: 12-19. [doi | wiley | pdf | supp.]

[3] Pausas JG & Bond WJ. 2022. Feedbacks in ecology and evolution. Trends Ecol Evol 37: 637-644. [doi | sciencedirect | pdf]

[4] Pausas JG. 2017. Bark thickness and fire regime: another twist. New Phytol 213: 13-15. [doi | wiley | pdf]

[5] Pausas JG & Keeley JE. 2017. Epicormic resprouting in fire-prone ecosystems. Trends Pl Sci 22: 1008-1015. [doi | sciencedirect | pdf

[6] Pausas JG, Lamont BB, Paula S, Appezzato-da-Glória B & Fidelis A. 2018. Unearthing belowground bud banks in fire-prone ecosystems. New Phytol 217: 1435–1448. [doi | pdf | suppl. | BBB database]

Fire regimes across the western Palearctic

July 20th, 2022 No comments

Fire regimes are shaped by climate, landscape structure, and the frequency of ignitions [1] and so globally vary across space, biogeographies, and environments. In a recent paper [3] we show how different fire regime parameters (e.g., area burnt, size, intensity, season, patchiness, pyrodiversity) varia across western Palearctic (Europe, North Africa, Near East) using remotely sensed data. We first divided the study area into eight large ecoregions based on their environment and vegetation: Mediterranean, Arid, Atlantic, Mountains, Boreal, Steppes, Continental, and Tundra. Then we characterize the fire regime for each region. The results show that the Mediterranean had the largest, most intense, and most recurrent fires, but the Steppes had the largest burnt area. Arid ecosystems had the most extended fire season, Tundra had the patchiest fires, and Boreal forests had the earliest fires of the year. The spatial variability in fire regimes was largely explained by the variability of climate and vegetation, with a tendency for greater fire activity in the warmer ecoregions. There was also a temporal tendency for fires to become larger during the last two decades, especially in Arid and Continental environments.

Figure 1. Fire size and fire intensity in eight ecoregions across western Palaearctic. From [3]
Fig. 2. Mean fire size (ha) and mean fire intensity (MW) in relation with Temperature of the driest quarter, for the eight ecoregions (colors as in Fig. 1 above). From [3].


[1] Pausas J.G. & Keeley J.E. 2021. Wildfires and global change. Front. Ecol. & Environ. 19: 387-395. [doi | wiley | pdf ]

[2] Pausas J.G. & Ribeiro E. 2013. The global fire-productivity relationship. Global Ecol. & Biogeogr. 22: 728-736. [doi | pdf | appendix | erratum ]

[3] Pausas J.G. 2022. Pyrogeography across the western Palearctic: a diversity of fire regimes. Global Ecol. & Biogeogr. [doi | wiley | pdf |data: dryad]

Feedbacks in ecology and evolution

April 21st, 2022 No comments

Ecology and evolutionary biology have focused on how organisms fit the environment. Less attention has been given to the idea that organisms can also modify their environment, and that these modifications can feed back to the organism, thus, providing a key factor for their persistence and evolution [1]. We propose that there are at least three independent lines of evidence emphasising these biological feedback processes at different scales (figure below): niche construction (population scale); alternative biome states (community scale); and the Gaia hypothesis (planetary scale). Flammability is an example of niche construction [2], and the forest-savanna mosaics are an example of the alternative biome states [3] (figure below). 

The importance of feedback processes make us rethink traditional concepts like niche and adaptation. For instance, the idea of evolution as a process of adaptation to fit a pre-existing environment needs to be replaced by a ‘co-evolutionary’ species-environment approach. An implication is that the concept of species niche, and niche occupancy, is less relevant than traditionally thought. That is, organisms do not adapt to a pre-existing environment (available niche), they construct their environment and then both ‘co-evolve’. A higher level of fitness is the result of this coevolution. Feedbacks also provide an alternative framework for understanding spatial and temporal patterns of vegetation that differ from those based on gradual changes (e.g., gradient analysis and succession), and suggest that multi-stability and abrupt transitions in a given environment are common [3]; this also has implications for species’ niche modelling [4].

Earth is in transition to a new and warmer state due to anthropogenic forcing, and feedback thinking may help us understand the process. We suggest that incorporating feedback thinking and understanding how feedbacks may operate at different scales may help in opening our minds to key processes contributing to the dynamics and resilience of our biosphere.

Fig. 1. Examples of eco-evolutionary feedbacks at different organising levels: Niche construction (population; e.g. flammability), alternative biome states (community; forests and savannas) and Gaia (biosphere). The signs of the feedbacks indicate the most common type of feedback for each example. Evolutionary feedbacks represent the evolutionary processes at the different scales (from selection at the micro-evolutionary scale to the acquisition of key macro-evolutionary innovations). From [1].


[1] Pausas J.G. & Bond W.J. 2022. Feedbacks in ecology and evolution. Trends Ecol. Evol. [doi | pdf]

[2] Pausas J.G., Keeley J.E., Schwilk D.W. 2017. Flammability as an ecological and evolutionary driver. J. Ecol. 105: 289-297. [doi | wiley | pdf]

[3] Pausas J.G. & Bond W.J. 2020. Alternative biome states in terrestrial ecosystems. Trends Plant Sci. 25: 250-263. [doi | sciencedirect | cell | pdf]

[4] Pausas J.G. & Bond W.J. 2021. Alternative biome states challenge the modelling of species’ niche shifts under climate change. J. Ecol. 109: 3962-3971. [doi | wiley | pdf]

Megafauna history and plant defense traits

January 10th, 2022 No comments

The role of large herbivores in explaining broad-scale ecological pattern has often been underestimated [1]. Plants have defenses against large herbivores (e.g., spines, high wood density [2]). And many continents had abundant large herbivores (megafauna) that were extinguished in Pleistocene (except in Africa). In a recent paper [3] we asked, to what extent the past distribution of extinct magafauna explains current geographical distribution of plant defense traits in the Neotropics (South & Central America). We fond that a significant proportion of the variance in the distribution of wood density, leaf size, stem spines, and leaf spines are explained by variable related to past megafauna (richness and body mass).

We defined 3 antiherbiomes in South America, that is, regions with characteristic plant defenses, environmental conditions, and Pleistocene megafauna, as follows: Small-Leaves-Thorny (SLT): thorny and small-leaved plants, in arid, cold and nutrient-rich ecosystems, containing numerous extinct and extant large grazers. Intermediate-Leaves-Woody (ILW): intermediate leaf sizes and levels of chemical defenses, and very high wood density, in moist and hot climates, and extremely nutrient-poor soils; and a high extinct megafauna richness, especially in relation to small browsers and mixed-feeders. Broad-Chemically-defended-Leaves (BCL): very large leaves with chemical defenses, mostly associated with moist climates and intermediate fertility soils, with few but large extinct megafauna species, especially browsers. Similar antiherbiomes can be observed in current Africa. These antiherbiomes represent one of the most striking broad-scale anachronisms in ecology.

We estimated that in South America, savannas occupied about 10 millions of Km2 during the Pleistocene, ca. 63% of them were converted to forests (44% to moist forests, 19% to dry forests) after the megafauna extinction (biome shifts [4]), and ca. 37% remains as savanna (stable). This suggests that South America was a savanna-dominated continent, much more similar to Africa than today, and that a large proportion of South American forests are the result of megafauna extinctions.

Overall our results suggest that past (extinct) large herbivores explain an important proportion of the variability of current plant traits and community assemblies.


Fig. 1. Left: Distribution of the 3 anti-herbiomes. Right: Hypothesized distribution of savanna during the Pleistocene (coloured areas; based on the distribution of extinct megafauna), that currently are savanna (in yellow), moist forests (dark green) and dry forests (light green). From [3]
Fig. 2. Reconstruction of Pleistocene savanna (ILW antiherbiome) with Taxodon platensis (a mixed feeder) next to the tree Bowdichia virgilioides (sucupira-preto; Fabaceae), and a Notiomastodon in the background. Artist: Júlia d’Oliveira

Fig. 3. Additional reconstructions of the Pleistocene Brazilian savannas from [5]. Artist: Júlia d’Oliveira



[1] Pausas JG & Bond WJ. 2019. Humboldt and the reinvention of nature. J. Ecol. 107: 1031-1037. [doi | jecol blog | jgp blog | pdf]

[2] Dantas V & Pausas JG. 2020. Megafauna biogeography explains plant functional trait variability in the tropics. Glob. Ecol. & Biogeogr. [doi | pdf | data:dryad | blog ]

[3] Dantas V., Pausas J.G. 2022. The legacy of Southern American extinct megafauna on plants and biomes. Nature Comm. 13: 129 [doi | pdf | data & codes] – New!

[4] [2] Pausas JG & Bond WJ. 2020. Alternative biome states in terrestrial ecosystems. Trends Plant Sci. 25: 250-263. [doi | sciencedirect | cell | pdf]

[5] Pansani et al. 2019. Isotopic paleoecology (δ13C, δ18O) of Late Quaternary megafauna from Mato Grosso do Sul and Bahia States, Brazil. Quat Sci Rev, 221, 105864. 

Reconciling Gleason’s and Clements’ views

September 30th, 2021 No comments

The question of whether species are organised as collectives of integrated interacting assemblages (Clements’ community concept) or behave individualistically (Gleason’s community concept) is a century-old debate in ecology that is still unresolved. In a recent article, we are reconciling the two approaches [1].

The Gleasonian view suggests that communities are assembled by species that respond individualistically along environmental gradients and thus cannot form bounded units (Fig. 1A). However, in many world landscapes, for a given climate, strikingly different biomes with sharp boundaries co-occur forming landscape mosaics. These mosaics are typically formed by a closed biome (forests) and open (non-forest) biome (e.g., grassland, savanna, shrublands). These two alternative biome states (ABSs [2]) are maintained by different feedback processes and have radically different species with contrasting shade and disturbance tolerance traits [2].

Under the individualistic view of species along climatic gradients, the overlapping response curve along a climate gradient (Fig. 1A) may indicate plant coexistence (and potentially competitive interactions); however this is true only if they occur in the same biome (Fig. 1B). That is both Gleason’s individualistic view (within biome) and Clements’s organismic view (across biomes) are complementary; both perspective of community remain useful in ecology.

The consequence is that fitting species distribution models or using climate limits in modelling for projecting future species distributions are inappropriate for extensive regions with alternative biome states. One way to improve these predictions would be to consider the presence or absence of forest shade in the modelling [1].

Figure 1. Classical (Gleasonian) pattern of species response curves along a climate gradient (A), and the alternative pattern along the same climatic gradient when there are ABSs (B). Note that in the driest and the wettest section of the gradient, we find open (e.g., grassland) and closed (forest) biomes, respectively; but at intermediate levels of the gradient, both are possible depending mainly on the disturbance regimes and feedback processes [2). Thus, under the intermediate levels of the gradient, species that may seem to coexist when considering climate only (A) are not really coexisting but occurring in drastically different biomes (B). From [1].


[1] Pausas J.G. & Bond W.J. 2021. Alternative biome states challenge the modelling of species’ niche shifts under climate change. J. Ecol. 109: 3962-3971 [doi | pdf]

[2] Pausas J.G. & Bond W.J. 2020. Alternative biome states in terrestrial ecosystems. Trends Pl. Sci. 25: 250-263. [doi | sciencedirect | cell | pdf]  

Alternative Biome States

January 8th, 2020 No comments

There is growing interest in the application of alternative stable state (ASS) theory to explain major vegetation patterns in tropical ecosystems [1] and beyond [2]. In a recent paper [3] we introduced the theory as applied to the puzzle of non-forested (open) biomes growing in climates that are warm and wet enough to support forests (alternative biome states, ABSs; Fig. 1). Long thought to be the product of deforestation, diverse lines of evidence indicate that many open ecosystems are ancient. They have also been characterized as ‘early successional’ even where they persist for millennia. ABS is an alternative framework to that of climate determinism and succession (Table 1 below) for exploring forest/nonforest mosaics. Within climatic and edaphic constraints, consumers (fire and herbivores) can produce vastly different ecosystems from the climate potential and have done so for millions of years [4]. This framework explains not only tropical forest–savanna landscapes, but also other landscape mosaics across the globe (Fig. 2).

Fig. 1. Generalized feedback processes in fire-prone landscapes where open and closed biomes (e.g., a grassland and forest) are alternative stable states maintained by stabilizing feedbacks, while perturbations generate abrupt transitions among states (destabilizing factors). From: [3].

Fig. 2. Examples of multibiome landscape mosaics where closed forests alternate with open biomes (grasslands) that are maintained by mammal herbivory and fire. From: [3].

Table 1. Comparison of the three main dynamic processes assembling disturbance-prone communities and landscapes: classical (facilitation) succession, autosuccession, and ABS. From: [3].


[1] Dantas V.L., Hirota M., Oliveira R.S., Pausas J.G. 2016. Disturbance maintains alternative biome states. Ecol. Lett. 19: 12-19. [doi | wiley | pdf | suppl.]

[2] Pausas, J.G. 2015. Alternative fire-driven vegetation states. J. Veget. Sci. 26:4-6. [doi | pdf | suppl.]

[3] Pausas J.G. & Bond W.J. 2020. Alternative biome states in terrestrial ecosystems. Trends Plant Sci. [doi | sciencedirect| pdf]

[4] Pausas J.G. & Bond W.J. 2019. Humboldt and the reinvention of nature. J. Ecol. 107: 1031-1037. [doi | jecol blog | jgp blog | pdf]