One (more) step towards a dynamic view of the Latitudinal Diversity Gradient
Ancient tropical extinctions contributed to the latitudinal diversity gradient
The Latitudinal Diversity Gradient (LDG) has fascinated natural historians, ecologists and evolutionary biologists ever since  described it about 200 years ago . Despite such interest, agreement on the origin and nature of this gradient has been elusive. Several tens of hypotheses and models have been put forward as explanations for the LDG [2-3], that can be grouped in ecological, evolutionary and historical explanations  (see also ). These explanations can be reduced to no less than 26 hypotheses, which account for variations in ecological limits for the establishment of progressively larger assemblages, diversification rates, and time for species accumulation . Besides that, although in general the tropics hold more species, different taxa show different shapes and rates of spatial variation , and a considerable number of groups show reverse patterns, with richer assemblages in cold temperate regions (see e.g. [7-9]).
Understanding such complexity needs integrating ecological and evolutionary research into the wide temporal and spatial perspectives provided by the burgeoning field of biogeography. This integrative discipline ¬–that traces back to Humboldt himself (e.g. )– seeks to put together historical and functional explanations to explain the complex dynamics of Earth’s biodiversity. Different to quantum physicists, biogeographers cannot pursue the ultimate principle behind the patterns we observe in nature due to the interplay of causes and effects, which in fact tell us that there is not such a single principle. Rather, they need to identify an array of basic principles coming from different perspectives, to then integrate them into models that provide realistic –but never simple– explanations to biodiversity gradients such as LDG (see, e.g., [5; 11]). That is, rather than searching for a sole explanation, research on the LDG must aim to identify as many signals hidden in the pattern as possible, and provide hypotheses or models that account for these signals. To later integrate them and, whenever possible, to validate them with empirical data on the organisms’ distribution, ecology and traits, phylogenies, fossils, etc.
Within this context, Meseguer & Condamine  provide a novel perspective to LDG research using phylogenetic and fossil evidence on the origin and extinction of taxa within the turtle, crocodile and lizard (i.e. squamate) lineages. By digging into deep time down to the Triassic (about 250 Myr ago) they are able to identify several episodes of flattening and steepening of the LDG for these three clades. Strikingly, their results show similar diversification rates in the northern hemisphere and in the equator during the over 100 Myr long global greenhouse period that extends from the late Jurassic to the Cretaceous and early Neogene. During this period, the LDG for these three groups would have appeared quite even across a mainly tropical Globe, although the equatorial regions were apparently much more evolutionarily dynamic. The equator shows much higher rates of origination and extinction of branches throughout the Cretaceous, but they counteract each other so net diversification is similar to that of the northern hemisphere in all three groups. The transition to a progressively colder Earth in the Paleogene (starting around 50 Myr ago) provokes a mass extinction in the three clades, which is compensated in the equator by the dispersal of many taxa from the areas that currently pertain to the Holarctic biogeographical realm. Finally, during the coldhouse Earth’s climatic conditions of the Neogene only squamates show significant positive diversification rates in extratropical areas, while the diversity of testudines remains, and crocodiles continue declining progressively towards oblivion in the whole world.
Meseguer & Condamine  attribute these temporal patterns to the so-called asymmetric gradient of extinction and dispersal (AGED) framework. Here, the dynamics of extinction-at and dispersal-from high latitudes during colder periods increase the steepness of the LDG. Whereas the gradient flattens when Earth warms up as a result of dispersal from the equator followed by increased diversification in extratropical regions. This idea in itself is not new, for the influence of climatic oscillations on diversification rates is well known, at least for the Pleistocene Ice Ages , as is the effect of niche conservatism on the LDG . Nevertheless, Meseguer & Condamine’s AGED provides a synthetic verbal model that could allow integrating the three main types of processes behind the LDG into a single framework. To do this it would be necessary to combine AGED’s cycles of dispersal and diversification with realistic models of: (1) the ecological limits to host rich assemblages in the colder and less productive temperate climatic domains; (2) the variations in diversification rates with shifts in temperature and/or energy regimes; and (3) the geographical patterns of climatic oscillation through time that determine the time for species accumulation in each region.
Integrating these models may allow transposing Meseguer & Condamine’s  framework into the more mechanistic macroecological models advocated by Pontarp et al. . This type of mechanistic models has been already used to understand the development of biodiversity gradients through the climatic oscillations of the Pleistocene and the Quaternary (e.g. ). So the challenge in this case would be to generate a realistic scenario of geographical dynamics that accounts for plate tectonics and long-term climatic oscillations. This is still a major gap and we would benefit from the integrated work by historical geologists and climatologists here. For instance, there is little doubt about the progressive cooling through the Cenozoic based in isotope recording in sea floor sediments . Meseguer & Condamine  use this evidence for separating greenhouse, transition and coldhouse world scenarios, which should not be a problem for these rough classes. However, a detailed study of the evolutionary correlation of true climate variables across the tree of life is still pending, as temperature is inferred only for sea water in an ice-free ocean, say earlier half of the Cenozoic . Precipitation regime is even less known. Such scenario would provide a scaffold upon which the temporal dynamics of several aspects of the generation and loss of biodiversity can be modelled. Additionally, one of the great advantages of selecting key clades to study the LDG would be to determine the functional basis of diversification. There are species traits that are well known to affect speciation and extinction probabilities, such as reproductive strategies or life histories (e.g. ). Whereas these traits might also be a somewhat redundant effect of climatic causes, they might foster (i.e. “extended reinforcement”, ) or slow diversification. Even so, it is unlikely that such a model would account for all the latitudinal variation in species richness. But it will at least provide a baseline for the main latitudinal variations in the diversity of the regional communities (sensu ) worldwide. Within this context the effects of recent ecological, evolutionary and historical processes, such as environmental heterogeneity, current diversification rates or glacial cycles, will only modify the general LDG pattern resulting from the main processes contained in Meseguer & Condamine’s AGED, thereby providing a more comprehensive understanding of the geographical gradients of diversity.
 Humboldt, A. v. (1808). Ansichten der Natur, mit wissenschaftlichen Erläuterungen. J. G. Cotta, Tübingen.
 Hawkins, B. A. (2001). Ecology's oldest pattern? Trends in Ecology & Evolution, 16, 470. doi: 10.1016/S0169-5347(01)02197-8
 Lomolino, M. V., Riddle, B. R. & Whittaker, R. J. (2017). Biogeography. Fifth Edition. Sinauer Associates, Inc., Sunderland, Massachussets.
 Mittelbach, G. G., Schemske, D. W., Cornell, H. V., Allen, A. P., Brown, J. M., Bush, M. B., Harrison, S. P., Hurlbert, A. H., Knowlton, N., Lessios, H. A., McCain, C. M., McCune, A. R., McDade, L. A., McPeek, M. A., Near, T. J., Price, T. D., Ricklefs, R. E., Roy, K., Sax, D. F., Schluter, D., Sobel, J. M. & Turelli, M. (2007). Evolution and the latitudinal diversity gradient: speciation, extinction and biogeography. Ecology Letters, 10, 315-331. doi: 10.1111/j.1461-0248.2007.01020.x
 Pontarp, M., Bunnefeld L., Cabral, J. S., Etienne, R. S., Fritz, S. A., Gillespie, R. Graham, C. H., Hagen, O., Hartig, F., Huang, S., Jansson, R., Maliet, O., Münkemüller, T., Pellissier, L., Rangel, T. F., Storch, D., Wiegand, T. & Hurlbert, A. H. (2019). The latitudinal diversity gradient: novel understanding through mechanistic eco-evolutionary models. Trends in ecology & evolution, 34, 211-223. doi: 10.1016/j.tree.2018.11.009
 Hillebrand, H. (2004). On the generality of the latitudinal diversity gradient. The American Naturalist, 163, 192-211. doi: 10.1086/381004
 Santos, A. M. C. & Quicke, D. L. J. (2011). Large-scale diversity patterns of parasitoid insects. Entomological Science, 14, 371-382. doi: 10.1111/j.1479-8298.2011.00481.x
 Morinière, J., Van Dam, M. H., Hawlitschek, O., Bergsten, J., Michat, M. C., Hendrich, L., Ribera, I., Toussaint, E. F. A. & Balke, M. (2016). Phylogenetic niche conservatism explains an inverse latitudinal diversity gradient in freshwater arthropods. Scientific Reports, 6, 26340. doi: 10.1038/srep26340
 Weiser, M. D., Swenson, N. G., Enquist, B. J., Michaletz, S. T., Waide, R. B., Zhou, J. & Kaspari, M. (2018). Taxonomic decomposition of the latitudinal gradient in species diversity of North American floras. Journal of Biogeography, 45, 418-428. doi: 10.1111/jbi.13131
 Humboldt, A. v. (1805). Essai sur la geographie des plantes; accompagné d'un tableau physique des régions equinoxiales. Levrault, Paris.
 Rangel, T. F., Edwards, N. R., Holden, P. B., Diniz-Filho, J. A. F., Gosling, W. D., Coelho, M. T. P., Cassemiro, F. A. S., Rahbek, C. & Colwell, R. K. (2018). Modeling the ecology and evolution of biodiversity: Biogeographical cradles, museums, and graves. Science, 361, eaar5452. doi: 10.1126/science.aar5452
 Meseguer, A. S. & Condamine, F. L. (2019). Ancient tropical extinctions contributed to the latitudinal diversity gradient. bioRxiv, 236646, ver. 4 peer-reviewed and recommended by PCI Evol Biol. doi: 10.1101/236646
 Jansson, R., & Dynesius, M. (2002). The fate of clades in a world of recurrent climatic change: Milankovitch oscillations and evolution. Annual review of ecology and systematics, 33(1), 741-777. doi: 10.1146/annurev.ecolsys.33.010802.150520
 Wiens, J. J., & Donoghue, M. J. (2004). Historical biogeography, ecology and species richness. Trends in ecology & evolution, 19, 639-644. doi: 10.1016/j.tree.2004.09.011
 Zachos, J. C., Dickens, G. R., & Zeebe, R. E. (2008). An early Cenozoic perspective on greenhouse warming and carbon-cycle dynamics. Nature, 451, 279-283. doi: 10.1038/nature06588
 Zúñiga-Vega, J. J., Fuentes-G, J. A., Ossip-Drahos, A. G., & Martins, E. P. (2016). Repeated evolution of viviparity in phrynosomatid lizards constrained interspecific diversification in some life-history traits. Biology letters, 12, 20160653. doi: 10.1098/rsbl.2016.0653
 Butlin, R. K., & Smadja, C. M. (2018). Coupling, reinforcement, and speciation. The American Naturalist, 191, 155-172. doi: 10.1086/695136
 Ricklefs, R. E. (2015). Intrinsic dynamics of the regional community. Ecology letters, 18, 497-503. doi: 10.1111/ele.12431
Joaquín Hortal and Juan Arroyo (2019) One (more) step towards a dynamic view of the Latitudinal Diversity Gradient. Peer Community in Evolutionary Biology, 100068. 10.24072/pci.evolbiol.100068
Revision round #32019-02-06
Decision round #3
First of all, please accept our apologies for the delay with this review. You present a novel and very bold hypothesis about (one of) the causes of latitudinal diversity gradients, and you do so in a quite complete and thorough way. So analysing all aspects of it has required some time, which together with the change of recommender and one of the busiest periods of the last years has done the rest. We are very sorry for that.
Briefly, the current version of your study is in great shape, but needs some final tweaking. Besides our own reading, two additional reviewers made an in-depth review of it, and they identify some key points that merit some revisions. Most of these revisions are of relatively minor importance, as they are directed to improve the clarity of the manuscript and, importantly, the presentation of your results and analyses. As stated above, you tackle a novel issue and you do it thoroughly, so it is not surprising that some bits remain obscure even though the main lines of your work are quite convincing (but see my cautionary note below). So please follow the advice made by the reviewers, and try to clarify and/or discuss the points they make in a way that (a) the reasons behind your methodological choices are clear for the readers; and (b) the uncertainty associated to your results and their and implications are communicated better.
With regard to (a) both reviewers provide some clues about where the text needs improvement and some clarifications. Perhaps the most significant of them is Joaquín Calatayud's concern #2. This may be solved by reporting the state node probabilities as supplementary material, and making a brief statement that, given their distribution, the uncertainty associated to their estimations does not affect the overall estimation of the number of times each type of dispersal event (i.e. equatorward/poleward) happened. However, if the actual results do not provide solid evidence when uncertainty in state node estimations is displayed (i.e. if it not self-evident that there is no significant effect of uncertainty after a visual review), then I would recommend that you follow Dr. Calatayud's recommendations for estimating the uncertainty in the dispersal event associated to each branch of the phylogeny.
Re (b) we urge you to follow the reviewers's advice, in particular Dr. Calatayud's point #1, which I believe will improve the fairness in the presentation of your results. But importantly, we think that at the very end your manuscript may overstate the importance it has for the current configuration of the Latitudinal Diversity Gradient for the two ancient clades of turtles and crocodiles, and other tropical-adapted clades. It is not that your do not present convincing evidence that the ancient loss of tropical diversity has an imprint on the current configuration of the gradient. But the way you word things in the discussion and conclusions give the impression that current configuration is the direct result of this process, thus giving a secondary importance to all the ulterior processes niche-conservatism, post-glacial dispersal, adaptations to dryer environments, etc. This is a matter of wording, of course, but the devil is in the details. We believe that the text will be more fair to the complexity of processes and factors that give rise to the current diversity gradient if you would simply include your evidence within the framework provided by current state-of-the-art in latitudinal diversity gradients.
Such current state-of-the-art may not be clear, but it comprises a lot of different pieces of evidences and perspectives. perhaps starting with the reviews of Jansson & Dynesius (Ann Rev Ecol Evol Syst 2003) and Willig et al. (Ann Rev Ecol Evol Syst 2003), and including the outcome of, e.g., Hawkins et al. (Ecology 2003); Wiens & Donoghue (TREE 2004); Ricklefs (Ecology 2006; Am Nat 2008; Ecol Lett 2011); Mittelbach et al. (Ecol Lett 2007); Svenning & Skov (Ecol Lett 2004); Allen & Gilooly (Ecol Lett 2006); Arita & Vázquez-Domínguez (Ecol Lett 2008); or (of course) Condamine et al. (Ecol Lett 2012); and ending with the (arguably) mixed evidence provided by Hanly et al (Am Nat 2017); Rabosky et al (Nature 2018) or Rangel et al. (Science 2018), among others. Here I’m not claiming that you have to be exhaustive, but rather to reflect that there are several causes of the gradient that all are operating over the pattern we see today. This is something that can perhaps be solved tweaking a few sentences here and there and/or adding one or two more in the conclusions, perhaps without even adding any citation. But without it we believe that the text does not provide a fair account of its true importance. Note that this does not reduce its importance.
Besides these (relatively) minor concerns, we believe that your manuscripts presents an overall solid work, both conceptually and analytically, that has the potential to have a significant effect on current debate on the origin(s) of the latitudinal diversity gradient. In fact, it is not every year (or even decade) that we see a truly novel addition to the latitudinal diversity gradient debate. We are looking forward to recommend a final version of this work.
Reviewed by Joaquin Calatayud, 2018-11-23 05:38
Reviewed by anonymous reviewer, 2018-11-23 05:42
Revision round #22018-10-05
Decision round #2
AE report on https://doi.org/10.1101/236646
I apologize for the slowness of this review; as I intimated in my first report, I find this large and ambitious study difficult to evaluate. The "AGE hypothesis" that this paper offers seems to be a formalization of ideas that have been around for a long time, making it important, but its placement in the literature problematic. There is no doubt that the combination of reconstructed phylogenetic and fossil tests is novel, and offers a deep time perspective on the Latitudinal Diversity Gradient (LDG).
The authors make several claims about the process leading to the present-day LDG using several different data streams for several different taxa, sometimes keeping the data streams separate, and sometimes combining them. The main red thread seems to be that reconstructed phylogenies can capture neither (i) high extinction (leading to "negative diversification") in Holarctic regions, nor (ii) "into the tropics" dispersal and range contractions, both occurring during the transition from Greenhouse to Coldhouse conditions between ˜50mya to ˜23 mya. Because of this, these two processes are potentially underappreciated drivers of the LDG.
The authors present PyRate analyses of fossil genera for the three groups, considering origination and extinction of lineages in Holarctic and Equatorial realms for Greenhouse, Transition, and Coldhouse periods (so, two rates for three times for three clades, or 18 separate rates, Figure 4. To summarize, turtles show the predicted decrease in origination and increase in extinctions during the Transition phase in the Holarctic leading to negative diversification there. Crocodiles show a decrease in origination rates in both realms in the Transition phase, but an increase in extinction in equatorial realm during the transition phase, and no change in the Holarctic; overall, there seems to be negative diversification in the Holarctic since the onset of the Transition phase. Squamates show net decreases in origination in both realms with the onset of the Transition phase, though more extreme in the equatorial realm, and no change in extinction rates in either realm during the Transition phase, and, finally, a decrease in extinction rates in the Coldhouse phase. These major results suggest that there are potentially strong global climate – origination and extinction links in all three groups, but only the Turtles show an overall pattern consistent with AGE. (I have italicized the contrary-to-expectation patterns in this paragraph.) My reading of the methods suggests that Figure 5 is just a reworking of Figure 4 that combines place and time in the Greenhouse phase to produce 12 rather than 18 rates.
The authors then consider range contractions and dispersals. Here I think there is a very useful message – the authors offer great examples where fossil-based constraints on ancestral lineage area of occupancy leads to very different scenarios of dispersal. As the authors succinctly put it, there is no need to formally compare model fits of ancestral occupancy and movement, since the fossils are clear evidence that a particular lineage was in fact where the fossil was. This, and the comparison of BiSSE and PyRate estimates of diversification, highlight clearly how geographically biased extinction can severely bias any sort of reconstruction on a tree – we "knew" that nonhomogeneity of process can cause all sorts of problems, but these empirical comparisons that make use of fossil data hit home. These two sets of comparisons, if presented clearly, will be read with care and interest by comparative biologists.
With regard to the fossil-based reconstructions (Figure 7), all three groups show evidence for rates of dispersal into the equator to be generally increasing through time and greater than those out of the equator, though with no discernable effect of the various temperature phases. Range extinctions show idiosyncratic patterns, again with no clear effect of the Transition or the Cold phase.
As you can see, my reading and re-reading of the paper does not have me seeing how the data are so consistent with AGE as the authors. Given PCI is a nascent and voluntary endeavor, it is likely best that this go to a journal that has expert reviewers in paleontology and comparative methods at hand to offer another opinion.
I have, however, made extensive comments and suggestions on the manuscript itself (as notes using Preview), which I hope will help the authors as they prepare the paper for submission to a major journal. I will send this file to PCI directly to pass on. Overall, the multiple comparisons among constant rate BiSSE, time-dependent BiSSE, Pyrate using both time and space and time+space = "environment" sampling, and constrained, semi-constrained and unconstrained DEC analyses for each of the three clades make for a study that risks getting lost in its many, contradictory tests. That would be a shame, as the general idea of being able to test deep time climate-driven negative diversification using some sort of combined test with trees and PyRate would be a major advance in our thinking about this "oldest pattern in ecology."
Sincerely, Arne Mooers
Reviewed by Arne Mooers, 2018-07-31 11:44
Revision round #12018-06-14
Decision round #1
Associate Editor's - Arne Mooers - Comments.
This is an intriguing study. I commend the authors for all the work they have done, and I echo the reviewers that this could be a qualitative contribution to the LDG literature. I very much liked the review and Table 1 (though I wish it had been annotated just a wee bit more).
The AGE set-up does indeed seem synthetic, but it is formalizing and extending a lot that we already knew (e.g. that the LDG is not static because equatorial conditions have waxed and waned across latitudes through deep time, and the LDG could be due to tropical forms dying out in a cooler world farther from the poles) does not come till very late in the set-up (lines 133-147). This is confusing, because it is not clear if we are being exposed to a strong or weak test of a theory or to a new theory altogether.
I am also a bit more cautious than the reviewers regarding the results. I did not see a comparison of standard BiSSE analyses (which mirror those by Rolland et al. PloS Biol. 2014) with an AGE-specific BiSSE analyses. Would this not be a relevant test of AGE? If not (perhaps because the authors feel all BiSSE analyses are suspect due to, e.g., the lack of fossil constraints on dispersal), then this needs to be clarified, and the BiSSE results recast as "unhelpful" – at the moment, I am not sure if I am to try to interpret them, or not. (And if not, does this mean I should discount the Rolland et al., results too?). The comparison of the unconstrained and constrained DEC analyses is useful to show that uninformed reconstructions may be biased, but I am not sure that this was tested (do we have a metric to evaluate bias?), nor am I convinced that reconstructions assuming the fossil record is biogeographically "complete" for these groups over these time scales (the HFC DEC) can be considered data. I would want a terrestrial vertebrate paleontologist to comment on this specifically.
I think the tests of the AGE suggested by Reviewer 2 would be very interesting, but these suggestions also highlight what may be a major issue with this manuscript – it seems to at once want to introduce a new framework and also test the predictions from that new framework, and also present a lot of very deep-time and large-bin PyRate and DEC reconstructions across three large datasets. I am not convinced one can do all this at once effectively. The fact that none of the three groups actually present patterns consistent with AGE also gives me pause (am I correct with this interpretation? – figures 6 and 7 seem the most relevant here, but the fact I have to ask should give pause).
I wonder if a paper that focused on, e.g. the Testudines as a test case might have higher impact. One could explore the AGE in detail with a view to presenting how one might properly test the AGE with these sorts of data. I leave this, and the reviewers comments, for the authors to consider as they revise and improve the paper.