Final Report Summary - MEDIATEMP (Is there a limit to biotic diversification? Insights from stochastic models of speciation and extinction)
Is the biodiversity we see today on Earth the exclusive result of stochastic (random) processes or does it also reveal the importance of evolutionary/ecological differences between species?
In this project, our overall aim was to understand how biodiversity patterns emerge through time and to establish whether current biodiversity patterns in real taxa have reached their equilibrium.
First, we explored exhaustively (Objective 1), via spatially-explicit individual-based computer simulations, the macroecological and phylogenetic patterns of biodiversity, produced from a variety of stochastic diversification models, inspired by the Neutral Theory of Biodiversity, where individuals of all species are identical in their propensity to give birth, die, disperse (locally or globally) and produce new species (according to simple partition rules).
Second, we measured (Objective 2) how quickly these biodiversity patterns approached their equilibrium and discovered for the first time that biodiversity patterns tend not to converge to equilibrium at the same time. Species richness took the least amount of time to approach equilibrium, followed in the first instance by species area relationships (SAR) and then by species abundance distributions (SAD). Phylogenetic patterns of biodiversity took so long to stabilize (hundreds of times longer than macroecological patterns) to make predictions at equilibrium irrelevant, as no system could be expected to remain constant for so long. We also showed that the difference in timing of equilibrium between species richness, SAR and SAD tended to increase, up to a non-trivial factor of six, the more uneven the partition of a parent species into its two daughter species was at speciation. Given that real biotas (at the global scale) typically may not have reached equilibrium in terms of species richness (based on fossil records or analyses of phylogenetic trees), our study successfully showed that predicting regional and global biodiversity patterns at equilibrium is unlikely to be appropriate (as is typically done in the current prevailing paradigm) to understand biodiversity.
Third, we focussed on understanding (Objective 3) how each biodiversity pattern changed as it approached equilibrium and to what extent it differed between stochastic scenarios, to better understand the respective influence of different stochastic processes.
In our simulations, the shape that SADs took were influenced primarily by speciation mode and very little (only incurring slight changes in parameter values) by dispersal mode and speciation rate. When the probability for a species to form new species was directly proportional to the species abundance (as modelled in neutral theory) the SAD recovered at equilibrium and large spatial scale invariably followed the Negative Binomial family of models (as expected), but when the probability of forming new species was the same for all species (irrespective of their abundance) the SAD followed instead a Poisson lognormal model. This distinction is meaningful because some authors have interpreted the overwhelmingly stronger support for the Poisson lognormal model compared to the Negative Binomial model as evidence that marine communities were not neutral and that species differences were a key driver of community structure. We propose instead that marine communities could still be neutral but that speciation probability is the same for all species.
Contrary to SADs, the shape that SARs took were influenced primarily by dispersal mode and less so by speciation rate and speciation mode. In order to obtain realistic SAR patterns at all spatial scales, our results confirm that finite dispersal (i.e. not panmictic) is essential.
Finally, in terms of phylogenetic structure, our project also shed some new light on these less well known biodiversity patterns in the context of neutral theory. A previous simulation study showed that neutral models were able to mimic realistic patterns of phylogenetic imbalance (a measure of how unevenly distributed species are among the tree branches) but were unable to produce realistic patterns of diversification through time (producing an acceleration of speciation events towards the present when real phylogenies show either a constant rate of diversification or a slowdown). Our study, however, showed that an intermediate mode of speciation (in between the two extreme modes studied before) was able to produce realistic patterns of both phylogenetic imbalance and diversification tempo, provided the system was still actively increasing in species richness (i.e. had not reached its equilibrium between speciations and extinctions). This striking result reopens the possibility that the neutral theory (at least a version of it) is a viable explanation for the phylogenetic patterns seen in nature and at the same time could suggest that the vast majority of organisms are either still actively increasing in species richness or have reached an equilibrium between speciations and extinctions only recently.
Overall, the conclusions of this project so far is that neutral theory may, after all, be able to account for and predict simultaneously the full range of biodiversity patterns (both macroecological and phylogenetic) seen in nature, if the right combination of stochastic neutral processes is chosen (finite dispersal, intermediate speciation mode and speciation probability being the same for all species) and one remains open to the possibility that biotas may not have reached equilibrium yet. None of this was known before we started our project and many researchers doubted that neutral theory could actually achieve this without adding some non-neutral processes.
We were also hoping (Objective 4) to develop new analytical tools and apply them on some well known biotas to identify the stochastic diversification scenario that most closely matched their biodiversity patterns and determine how far they are from reaching equilibrium. Alas, we have not made much progress on this front yet. We are still actively pursuing this objective, however, and feel confident that we will find a working analytical solution soon.
In this project, our overall aim was to understand how biodiversity patterns emerge through time and to establish whether current biodiversity patterns in real taxa have reached their equilibrium.
First, we explored exhaustively (Objective 1), via spatially-explicit individual-based computer simulations, the macroecological and phylogenetic patterns of biodiversity, produced from a variety of stochastic diversification models, inspired by the Neutral Theory of Biodiversity, where individuals of all species are identical in their propensity to give birth, die, disperse (locally or globally) and produce new species (according to simple partition rules).
Second, we measured (Objective 2) how quickly these biodiversity patterns approached their equilibrium and discovered for the first time that biodiversity patterns tend not to converge to equilibrium at the same time. Species richness took the least amount of time to approach equilibrium, followed in the first instance by species area relationships (SAR) and then by species abundance distributions (SAD). Phylogenetic patterns of biodiversity took so long to stabilize (hundreds of times longer than macroecological patterns) to make predictions at equilibrium irrelevant, as no system could be expected to remain constant for so long. We also showed that the difference in timing of equilibrium between species richness, SAR and SAD tended to increase, up to a non-trivial factor of six, the more uneven the partition of a parent species into its two daughter species was at speciation. Given that real biotas (at the global scale) typically may not have reached equilibrium in terms of species richness (based on fossil records or analyses of phylogenetic trees), our study successfully showed that predicting regional and global biodiversity patterns at equilibrium is unlikely to be appropriate (as is typically done in the current prevailing paradigm) to understand biodiversity.
Third, we focussed on understanding (Objective 3) how each biodiversity pattern changed as it approached equilibrium and to what extent it differed between stochastic scenarios, to better understand the respective influence of different stochastic processes.
In our simulations, the shape that SADs took were influenced primarily by speciation mode and very little (only incurring slight changes in parameter values) by dispersal mode and speciation rate. When the probability for a species to form new species was directly proportional to the species abundance (as modelled in neutral theory) the SAD recovered at equilibrium and large spatial scale invariably followed the Negative Binomial family of models (as expected), but when the probability of forming new species was the same for all species (irrespective of their abundance) the SAD followed instead a Poisson lognormal model. This distinction is meaningful because some authors have interpreted the overwhelmingly stronger support for the Poisson lognormal model compared to the Negative Binomial model as evidence that marine communities were not neutral and that species differences were a key driver of community structure. We propose instead that marine communities could still be neutral but that speciation probability is the same for all species.
Contrary to SADs, the shape that SARs took were influenced primarily by dispersal mode and less so by speciation rate and speciation mode. In order to obtain realistic SAR patterns at all spatial scales, our results confirm that finite dispersal (i.e. not panmictic) is essential.
Finally, in terms of phylogenetic structure, our project also shed some new light on these less well known biodiversity patterns in the context of neutral theory. A previous simulation study showed that neutral models were able to mimic realistic patterns of phylogenetic imbalance (a measure of how unevenly distributed species are among the tree branches) but were unable to produce realistic patterns of diversification through time (producing an acceleration of speciation events towards the present when real phylogenies show either a constant rate of diversification or a slowdown). Our study, however, showed that an intermediate mode of speciation (in between the two extreme modes studied before) was able to produce realistic patterns of both phylogenetic imbalance and diversification tempo, provided the system was still actively increasing in species richness (i.e. had not reached its equilibrium between speciations and extinctions). This striking result reopens the possibility that the neutral theory (at least a version of it) is a viable explanation for the phylogenetic patterns seen in nature and at the same time could suggest that the vast majority of organisms are either still actively increasing in species richness or have reached an equilibrium between speciations and extinctions only recently.
Overall, the conclusions of this project so far is that neutral theory may, after all, be able to account for and predict simultaneously the full range of biodiversity patterns (both macroecological and phylogenetic) seen in nature, if the right combination of stochastic neutral processes is chosen (finite dispersal, intermediate speciation mode and speciation probability being the same for all species) and one remains open to the possibility that biotas may not have reached equilibrium yet. None of this was known before we started our project and many researchers doubted that neutral theory could actually achieve this without adding some non-neutral processes.
We were also hoping (Objective 4) to develop new analytical tools and apply them on some well known biotas to identify the stochastic diversification scenario that most closely matched their biodiversity patterns and determine how far they are from reaching equilibrium. Alas, we have not made much progress on this front yet. We are still actively pursuing this objective, however, and feel confident that we will find a working analytical solution soon.