Why is extinction necessary for speciation

The most direct expectation of weak directionality is that clades should become rich in specialised species as time passes. Hence, under this theory, we predict that most species in the late phase of clade evolution should have traits typical of specialists, such as small range size and high degree of sympatry 17 , 18 , 29 , Given high degree of sympatry and reduced range size are presumed to depress diversification, according to this model there should be some point in time where both the regime of range size evolution and the diversification process shift Fig.

We tested this hypothesis by locating statistically significant shift points in the total range size, degree of sympatry, and net diversification rate curves. The data included 14, species and 84, fossil occurrences, spanning around million years from the early Cambrian trilobites and brachiopods, to late Cretaceous ammonites. For all of the analysed clades, we first computed the range size of each species per time bin, and the range size of the entire clade per time bin, which represents the union of individual species ranges Fig.

The use of cumulative range values, rather than time bin data, is appropriate as it smooth off unequal sampling and allows calculating effectively changes in the regime of geographical evolution of clade see below.

The slope of total range curve plotted versus time indicates the velocity of range size accumulation at the level of clade. It is equivalent to the average size of species ranges times species richness cumulated over all time bins. The total range curve is best fitted by either sigmoid, or generalized logistic curves, while the linear model is rejected for all the examined clades. This indicates that the increase in total range size slows down towards the recent, according to saturation dynamic Table 1.

To quantify the degree of sympatry Figs 1 B and 2 , we started by summing the geographic range of the entire clade over consecutive time bins. Then, we computed the area between total and clade range curves per unit time, under the specific hypothesis that the area difference between the two curves should be larger after the shift point, thereby indicating a higher degree of sympatry since Fig.

Eventually, we tested how often the difference between the two curves tends to increase after the shift points across clades, in keeping with our hypothesis 2, by means of the binomial distribution. The total geographic range is computed by summing the range of each species in each time interval and then over successive intervals.

In the figure, the shaded areas represent the species ranges. Species are indicated by capital letters. For each time bin, extinct species are indicated in grey color, living species are reported in black. Upper row: computation of the total range curve.

Lower row: computation of the clade range curve. Finally, we computed a second measure of the degree of sympatry at the level of bin. For each such bin, we took the ratio between the total range size summed algebraically over all species in the bin and the clade range size in the focal interval. This ratio represents the degree of overlap among individual species ranges. Shift points in degree of sympatry among species, total range size and net diversification rate are statistically closer in time to each other than expected by chance in 20 out of 30 cases, and for 16 out of 21 clades Table 2.

Both figures are statistically different from chance according to the binomial distribution Table 3 , indicating that the existence and temporal coincidence of shift points are robust. We then took the average ages of the three shifts to get a single shift point, and tested hypotheses 2 and 3.

In keeping with our predictions hypothesis 2 the degree of sympatry increases after 28 out of 30 shift points. The increase is temporally coincident with shift points Table 3. Then, we tested the prediction that species average range size decreases after the shift points hypothesis 3 , by dividing the total cumulative range curve by the number of species present in each time bin.

This allows testing the evolution of species average range size over time Fig. After the shift points this curve has a slope significantly different from zero fifteen times, 12 of them being negative i.

This is consistent with the idea that species after the shifts tend to be small-ranged and therefore specialists. Net diversification rates decrease after the shift points in 28 out of 30 cases and for all of the clades, thus supporting our predictions. Finally, as expected the extinction rates are positively correlated with the degree of sympatry in 12 clades out of 21, and the speciation rate is negatively correlated to sympatry in 13 clades Table 4.

Overall, 16 clades out of 21 show either decreased speciation or increased extinction as the degree of sympatry increases, in keeping with our prediction that sympatry as a consequence of specialisation depresses diversification to drive the clade extinct.

Overall, our results indicate that the distinction between an early and a late phase of clade evolution is useful, that the net diversification rate decreases consistently during the late phase, and that mainly specialist species, having high degree of range overlap to each other sympatry , make up the majority of clade biodiversity after the shift points.

We repeated all of the analyses excluding species with less than 10 total occurrences, in order to rule out the possibility that what we perceive as rarity, is in fact lack of preservation. The results are available as supplementary information. On such a reduced dataset, we located 22 shift points for seventeen clades. The shift points are statistically significant in 19 out of 22 cases Yet, the degree of sympatry after the shift is higher than before only 9 times, and the average range size is significant and negative just one time.

Taken at face value, these latter results are not supportive of weak directionality. Yet, it must be noted that by excluding rare or otherwise poorly sampled species from the dataset, we effectively removed those species whose effect on clade range size evolution we were seeking to test. Weak directionality theory derives from the idea that natural selection incessantly fine-tunes species to their environment.

Its payoff is evident in the short run 13 , 14 , 16 but over long periods of time, it acts as an evolutionary trap 9. While this has long being noted when dealing with the fate of individual species, the impact of such a trap on the history of clades is massive. We show that the degree of geographic range overlap among species increases as clades grow older and most specifically after the shiftpoints. In modern ecosystems, it was shown that niche specialisation promotes sympatry 17 , 18 , Hence, this result points to a non-random pattern of the incidence of specialized species over time.

We proved a direct link between sympatry and diversification, which taken together with the notion that sympatry tends to grow over time during the history of clades, is further, crucial evidence, that specialisation undermines clade survival in the long run. We expected sympatry to impact negatively on speciation rate, and positively on extinction rate, but only during the late phase of clade evolution i.

In this regard, it is especially interesting to note that we found sympatry to correlate negatively to diversification rate 12 times, and positively only 3 times. As per speciation rate, we retrieved 13 negative correlations to the degree of sympatry, and only 4 positive correlations. Overall, this suggest that sympatry alone could be enough to explain the common observation that origination rate decrease over time for purely biological reason, even when diversity is low 4 , 31 , The slow-down in geographic expansion that clades experience over time suggests that this mechanism is especially important when the available space for allopatric speciation becomes limited 33 , which our data suggest should happen after the shift points.

Obviously, the net diversification rate decreases towards the present in all of the clades we analysed. Yet, what is important is that we found this decline is temporally coincident with the increased level of sympatry. Patterns of decreased diversification towards the present have also been frequently inferred from phylogenies 34 , 35 , 36 and explained in that context as the product of limited opportunities for speciation as clade history progresses towards the present 37 , However, our data suggest that increased habitat specialisation, hence higher sympatry are fundamental aspects of such trend in diversification.

Our model is inherently simple and based on ideas that have roamed through the scientific literature for decades. We believe that the main obstacles to their further development until now have been the difficulties of estimating changes in clade range size from fossil data, the lack of adequate fossil databases, and a certain reluctance to use any concept of progress i.

The landscape metaphor emphasizes the trend towards higher-fitness morphologies and genotypes as drivers of weak directionality. Specialisation has often being envisaged as an evolutionary trap 9 but its potential contribution to clade extinction has been largely neglected since Cope first identified the link between specialisation and clade demise in his law of the unspecialised. Liow 41 found that long-lived crinoid taxa are less specialised in morphology than more derived types but see ref.

This bolsters the idea that specialisation increases extinction risk, and would support our interpretation here if such long-lived taxa are also stratigraphically older.

Evidence for such a pattern comes from a study of ours on placental mammals 14 where we demonstrated that specialised types are geologically younger than their relatives. Statistical modelling of the evolution of specialisation further support the idea that natural selection on local adaptation and habitat choice always leads to specialists, implying the latter appear later, on average, than generalists We will not push our model so far as to say that all of the clades have to follow the path it predicts.

According to the weak directionality theory, after the shift point the geographic evolution of clades proceeds towards a condition where a few, highly-sympatric, species coexist within a relatively small range. This is especially robust considering that a number of clades went extinct during a mass extinction, which means their natural course towards extinction was abruptly interrupted by something unrelated to range size dynamics within clade of any other biological attribute 10 , and the effect that specialisation had upon them.

Although this dynamic is true for most of the clades we analysed - rhynconelliform brachiopods Strophomenida, Spiriferinida, Orthotedida, and Productida , tabulate corals Favositida , stenolaemate bryozoans Fenestrida, Cystoporida, Trepostomida, Rhabdomesida , pteriacean bivalves Pterineidae , proetid trilobites Proetida , murchinsonid Lophospiridae , eogastropod Euomphalidae and bellerophontid gastropods Bellerophontida , and ammonoid ammonites Desmoceratida - there are also clades whose total range size keeps growing after the shift points, since the number of species remains high then - Auloporida tabulate corals , Rhabdomesida stenolaemata bryozoans , Strophomenida and Orthotetida rhynconellid brachiopods.

Not surprisingly, these are all clades whose species richness was still high in the late phase of their evolution see supplementary material for the geographic and net diversification rate paths of individual clades.

Indeed, the range size frequency distribution is usually right-skewed within clades This means that relatively large-ranged species are always expected to occur within a clade. In addition, after the shift point the decrease in species diversity implies that surviving species probably have the opportunity to occupy the range they left vacated.

If the clade range size does not decrease in the latter bins, such dynamics would imply that average range size would not decrease in many cases. Thus, expectation of reduced average range size for species over time is too simplistic Most of the clades mentioned so far conform to the predictions of the weak directionality model: they show high levels of sympatry, small average range size, and small total range size after the shift points. A manifold of taxa, though, show small total ranges, small average range and small degree of sympatry.

These clades survived what according to the model should have been their final extinction moment because of species having disjunctive ranges. Thus, although rare, there are cases of clades whose species become rare overall in their former ranges, rather than becoming restricted to sympatry in a small residual range. Although potential artifacts may result in waxing and waning of species ranges and diversity at the clade level 46 , our empirical data sets differ from the predictions of random models of clade evolution, in conforming to the predictions of higher levels of sympatry, hence high incidence of specialization, occurs during the late portion of a clade existence.

Weak directionality theory provides a consistent and widely applicable explanation for the extinction of clades. They diversify and survive because individual species become more and more specialised over time. Yet, the very reason for clade success also contains the seed of their decline. The path to extinction is neither simple nor monotonic. We identified more shift points than clades, implying that clades may survive moments of crisis.

It will be interesting to know how phenotypic diversity evolves during clade existence, in order to understand whether phylogenetic or developmental constraints on evolvability might prevent clades from escaping the evolutionary trap that weak directionality represents 47 , Each data point includes the paleocoordinates and the estimated minimum and maximum age of the fossil localities. Data pertain to five different animal phyla Arthropoda, Brachiopoda, Bryozoa, Cnidaria, Mollusca and cover some million years of the fossil record.

Overall, the database includes ca. The fossil record of individual clades was divided in equal-length time bins. The length of such intervals was clade-specific, meaning that we applied bins of different lengths as to maintain as many bins as possible while avoiding producing bins containing less than three species with at least three occurrences per species, which is the minimum requirement to calculate a species range size estimate.

We further removed the species and genera that lack a continuous stratigraphic range and dubbed with uncertain taxonomic classification e. After applying these selection criteria, we were left with 21 clades, including 14, species and 84, occurrences overall. In particular, when we had species geographic extent ranging less than decimal degrees in longitude, we used the Lambert Azimuthal Equal Area projection.

If the longitudinal extent exceeded decimal degrees and latitudinal extent exceeded 60 decimal degrees north or south to the equator, we then used the Albers Equal Area projection.

In the very special cases not considered in the above criteria we splitted the polygons in order to have different regions to be projected by means of Lambert Azimuthal Equal Area projection and then summed the areas computed for every single polygon portion to get the original species polygon area.

After computing individual species ranges, we summed them to get the total range in the bin. Time-bin total species ranges were then summed over consecutive bins to get the total range curve Fig.

For each bin and clade, we also computed the clade range the range effectively occupied by the entire clade in the focal bin, Fig. Clade ranges were summed over consecutive bins as well, to get the clade range curve Fig. The difference between the actual and the total ranges per bin depends on how much species ranges overlap to each other i. Finally, we divided the total range per bin by the number of species present in that bin. This gives the average species range size Fig.

We analyzed fossil occurrence data for each clade using the program PyRate 51 , 52 , which provides a joint Bayesian estimation of the preservation process and diversification dynamics.

We modeled fossil preservation by a non-homogeneous Poisson process with rate estimated from the data and expressing the mean number of expected occurrences per lineage per time unit in this case 1 Myr.

Under the PyRate framework, the preservation process is used to infer times of origination and extinction when applicable of all lineages, which represent the result of an unknown underlying birth-death process. Based on the estimated number of shifts and their temporal placement, we obtained marginal posterior distribution of speciation, extinction, and net diversification rates calculated within 1 Myr time bins through the lifespan of each clade. Shiftpoints estimation and testing: We identified shift points in time by using the cross-entropy method, which applies the Bayesian information criterion to locate significant changes in the trend along the net diversification, total range, and clade range curves Then, we computed the mean age distance among the three.

We compared such mean distance to a family of random mean age distances, to test whether the sum of the time distances among them was smaller than expected by chance, which would imply the shift points are statistically coincident in time. To perform such comparison, we sampled at random 9, times two ages from time-bins midpoints the degree of sympatry and cumulative total range curves are computed per time-bin , and one age from the net diversification rate, 1-my long sample.

Then, we calculated the mean distance among the three random time points. Eventually, we counted how many times the random mean time distance is higher than the real distance between breakpoints, to calculate the p-value for the hypothesis that real time distances are smaller than random means. Testing the increase in degree of sympatry after the shift point: After locating the shift points, we first tested whether the degree of sympatry increases after them the area test, see Fig.

To this aim, we computed the area difference i. Then, we divided each area differences to the corresponding lengths of time. This gives two area differences per unit time. Testing for average range size reduction after the shift point: According to weak directionality theory, the species average range size should decrease after the shift points.

To test such a prediction, we regressed species average range size against time, selecting only those time bins more recent than the shift points. A significant and negative regression slope would indicate the expected trend towards range size reduction slope test, see Fig.

Testing for the correlation between diversification and the degree of sympatry: We used a time-varying birth-death model to investigate the intensity and significance of correlations between speciation and extinction rates and the degree of sympatry UC Berkeley. At the most basic level, mass extinctions reduce diversity by killing off specific lineages , and with them, any descendent species they might have given rise to. In this way, mass extinction prunes whole branches off the tree of life.

But mass extinction can also play a creative role in evolution, stimulating the growth of other branches. The sudden disappearance of plants and animals that occupy a specific habitat creates new opportunities for surviving species. Over many generations of natural selection , these lineages and their descendent lineages may evolve specializations suited to the newly freed up resources and may take over ecological roles previously held by other species, or may evolve brand new ecological strategies.

In this way, mass extinction can level the evolutionary playing field for a brief time, allowing lineages that were formerly minor players to diversify and become more prevalent. The period ended with extinction of the dinosaurs and the rise of mammals. The audio, illustrations, photos, and videos are credited beneath the media asset, except for promotional images, which generally link to another page that contains the media credit.

The Rights Holder for media is the person or group credited. Tyson Brown, National Geographic Society. National Geographic Society. For information on user permissions, please read our Terms of Service. If you have questions about how to cite anything on our website in your project or classroom presentation, please contact your teacher.

They will best know the preferred format. When you reach out to them, you will need the page title, URL, and the date you accessed the resource. If a media asset is downloadable, a download button appears in the corner of the media viewer.

If no button appears, you cannot download or save the media. Text on this page is printable and can be used according to our Terms of Service. Any interactives on this page can only be played while you are visiting our website. You cannot download interactives. Extinction is the complete disappearance of a species from Earth.

Species go extinct every year, but historically the average rate of extinction has been very slow with a few exceptions. The fossil record reveals five uniquely large mass extinction events during which significant events such as asteroid strikes and volcanic eruptions caused widespread extinctions over relatively short periods of time.

Some scientists think we might have entered our sixth mass extinction event driven largely by human activity. Our planet is dependent on an interconnected system. If we lose one species, how does that impact the whole system? What if we lose hundreds? Help your students understand the gravity of extinction with these classroom resources. In the mids, Charles Darwin famously described variation in the anatomy of finches from the Galapagos Islands.

Alfred Russel Wallace noted the similarities and differences between nearby species and those separated by natural boundaries in the Amazon and Indonesia. Independently they came to the same conclusion: over generations, natural selection of inherited traits could give rise to new species.