A post on one of biology’s most confounding riddles: the latitudinal gradient in biodiversity.

A beautiful, but comparatively species poor forest in eastern Oregon

Explaining global patterns of biodiversity is a fundamental goal in biology. Understanding how the tens of millions of species on earth have arranged themselves into populations, communities, and ecosystems, is critical for conserving them in the face of a rapidly growing human population and global climate change.

ResearchBlogging.orgThe latitudinal gradient in species diversity is perhaps the most famous such pattern, and it has confounded biologists for decades. Almost invariably across taxonomic groups, hemispheres and continents, as one moves from polar regions towards the equator, species diversity increases (see the figure for a depiction of global bird diversity). The concept of diversity here can be broken down into three parts: “alpha diversity” or the diversity of species in a single location; “beta diversity”, or the turnover of species observed when moving among locations; and “gamma diversity” or the diversity of species found in an entire region. The latitudinal diversity gradient holds true for all three elements.

A recent paper by Kraft et al. (2011) takes an intriguing look at the cause of the beta portion of the latitudinal gradient in diversity. It has long been thought by biologists that the higher turnover in species diversity among sites in tropical versus temperate forests was an indicator that fundamentally different processes were at work in structuring them on a local scale. For example, tropical species might generally have more fine tuned habitat preferences, or have more limited dispersal abilities than temperate species. Either would likely lead to more rapid species turnover among sites. A previously untested alternative is that the increase in beta diversity is a simple consequence of an increase in regional species diversity (or gamma diversity) and requires no special explanation.

Kraft et al. tested this idea by examining patterns of woody plant species diversity along two transects. The first is a latitudinal gradient spanning over 100 degrees*, and the second an elevational gradient of 2250m (elevational and latitudinal gradients in diversity show remarkably similar patterns, but that is a story for another time…). Each site in a transect consists of ten 0.01 hectare plots for which the authors obtained alpha diversity. Gamma diversity is then the total diversity of all plots within a site, and beta diversity the turnover among those plots. They then generated the expected distribution of beta diversity for each site in the transect under the hypothesis that no special biological process is at work in the tropics. They did this by repeatedy randomly reshuffling their data for each site in a way that maintained site gamma diversity and relative species abundances**.

To the authors surprise, they could not reject this model. They found consistently higher beta diversity than expected, but found it across all sites in each transect. Deviations from expectation were not in any way correlated with latitude or elevation. In short, it appears that the gradient in beta diversity is the expected consequence of changes in gamma diversity. This has at least one major implication: that the local processes that work to assemble plant diversity in a tropical rainforest, an eastern deciduous forest, or a boreal spruce-fir forest may be much the same, they just have different starting material (gamma diversity) from which to work. This suggests that to find the ultimate causes of higher tropical diversity we might look to large scale, long term processes such as speciation, extinction and dispersal that govern gamma diversity. In other words (yeah, I’ll just go ahead and do it) nothing in biology makes sense except in the light of evolution.

*As an aside, one of the cool things about this paper is that the latitudinal transect data represents a novel use of an old dataset originally collected by Alwyn H. Gentry (a by-all-accounts brilliant botanist who died in a tragic plane crash in 1993). This data has been made available to researchers by the Missouri Botanical Garden and has now been cited over 114 times. As an evolutionary biologist I’m fairly used to seeing DNA sequence data used and reused (most journals require its deposition in a public database as a condition for publication) but it seems pretty unusual for an ecological dataset to have such a wide impact.

**Correction: An earlier version of this post stated that Kraft et al.’s null model maintained plot alpha diversities. It does not.



Kraft, N., Comita, L., Chase, J., Sanders, N., Swenson, N., Crist, T., Stegen, J., Vellend, M., Boyle, B., Anderson, M., Cornell, H., Davies, K., Freestone, A., Inouye, B., Harrison, S., & Myers, J. (2011). Disentangling the drivers of beta diversity along latitudinal and elevational gradients. Science, 333 (6050), 1755-8 DOI: 10.1126/science.1208584

Thomas, G., Orme, C., Davies, R., Olson, V., Bennett, P., Gaston, K., Owens, I., & Blackburn, T. (2008). Regional variation in the historical components of global avian species richness Global Ecology and Biogeography, 17 (3), 340-51 DOI: 10.1111/j.1466-8238.2008.00384.x

9 comments on “A post on one of biology’s most confounding riddles: the latitudinal gradient in biodiversity.

  1. And now you have me trying to think of another massively-cited ecological data set. Best I can come up with is the Worldclim/Bioclim geographic climate data, but that’s only because I happen to be playing around with it right now.

  2. noahmattoon says:

    ha. I suppose I didn’t really consider the existence of massive government- or NGO-collected data sets when I made that statement. so perhaps it’s not exactly accurate…

  3. TGIQ says:

    That Kraft et al. ref needs to go on my reading list for the week…thanks!

  4. Cool blog post Noah! I am actually working on doing a similar analysis using data from Bolivia, but comparing their small scale analyses to analyses at a larger spatial scale (where local assemblages are defined as 0.1ha plots, not 0.01ha, and plots are separated by many km, not contiguous). Results shall be published soon (I hope!). One detail that is important is that their null model did not maintained alpha diversity, just gamma.

  5. This picked up a potentially interesting comment over at Denim and Tweed, too.

  6. Elmer Rich says:

    There is the idea that this relates to parasite load. The higher the parasite load the more species developed as a counter measure.

  7. noahmattoon says:

    Hey Sebas,

    Thanks for the correction! I look forward to checking out your new work. Is it shaping up to be consistent with this paper? The comment Jeremy links to at Denim and Tweed also raises the issue of spatial scale…

  8. Hey Noah,
    Well, the results that I have so far confirm Kraft et al.’s analyses at their scale: you go from a strong elevation gradient to no gradient after using the null model. However, when increasing the scale of analysis, if anything, the null-model-corrected gradient in b diversity is even stronger (with a big peak at mid-elevations!). Need to think more about how to explain that…

Comments are closed.