Canopy interception and the dispersed puddle

Taking a walk through grass after a light rain is a soaking affair. Even walking through a recently mowed lawn in the morning would wet your sneakers while going to school. It was always better to let the sun come out for a little bit before short-cutting across a yard.

The principle that most children learn at a young age likely has important ramifications for understanding the dynamics of how grasslands work. Through one of two mechanisms, my guess is that canopy interception sets up a negative feedback loop that constrains how much grass is produced.

First, a quick review.

In grasslands, approximately half of the precipitation can be intercepted by biomass without reaching the soils. For small precipitation events, 70% of the precipitation can be intercepted by a dry canopy, with the fraction of precipitation intercepted declining with event size (Ataroff and Naranjo 2009). A single square meter of grassland can withhold 2 L of water from reaching the soil.

Relationship between precipitation and canopy interception for a tropical pasture grass. From Ataroff and Naranjo 2009.

Half of the precipitation that could fall on a grassland might never reach the soil. And the more grass there is, the less precipitation would reach the soil. Seasonally, as grass grows and canopies develop, the demand for water would be ever increasing. Yet, because of interception, less and less precipitation would reach the soil.

Increasing demand, decreasing supply. A classic negative feedback that would be limiting growth. Even if plants had access to deep water, the consequences might be greater for N supply and cause transitive limitation as surface soils where N mineralization occurs would be prevented from rewetting.

Evolutionarily, we haven't explored whether there would be selection on herbaceous species to promote (or not promote!) throughfall of precipitation. Altered leaf angle, waxy cuticles, stemminess, would all alter how much water is retained or passed on to the soil. Ecologically, with just a few papers on the topic, there are likely some large unexplored ramifications besides promoting seasonal water limitations. For example, from first principles, rain coming in larger events should promote growth, not retard it, as the water is stored in the soil rather than the canopy.

Most importantly of all, if you haven't learned it yet, never cut across a wet lawn in the morning wearing sneakers. Might as well jump in a puddle.

The nuts and bolts of transitive limitation

Patterns of soil moisture in the lowlands of an annually burned watershed at Konza Prairie. Soil moisture is expressed on a relative basis at 6 depths for 1993 (wet year) and 1994 (dry year).

Earlier, I had discussed a potentially interesting case of transitive limitation, i.e. when the low availability of one resource reduces the availability of another. In the case of water and nitrogen, it is unclear in grasslands whether the limitation ascribed to water could actually be due to low N availability. N mineralization is known to decrease with decreasing soil moisture. As such, as soil moisture declines, so should N mineralization.

The correlation between soil moisture and N mineralization does not necessarily mean that the two should co-limit across a range of soil moistures. In a given soil profile, soil organic N is generally concentrated in shallow depths, while soil moisture is more evenly distributed throughout the soil profile, if not greater at depth. As such, plants can have access to plenty of water at depth even if shallow soils have dried out. Soil N mineralization and moisture might be correlated for a given volume of soil, but not over the whole soil profile.

Konza is an interesting example. At different times, productivity is said to be limited by water and nitrogen, but the two have never been rectified. Do they simultaneously limit production? Does limitation vary over the course of a season, or across years? Or is it transitive?

If it is transitive, disentangling the two is not easy. Standard factorial resource addition experiments do not work since adding water would also increase N availability. Is there a way to add water without increasing N? Not easily from above. But you could add it from below.

Inferentially, if you look at Konza soil moisture patterns, there is always plenty of water at depth, even in dry times. In the above example, in 1994, soil moistures are depleted in shallow soils, but there is very little draw down of deep soils. Proximally, this could be due to the lack of roots at depth, but we are only talking 1 m. The dominant species could easily produce roots at 1 m--if there was a benefit to doing so. If productivity was water limited, there would be a benefit. Yet, if productivity was actually N limited, accessing deep water provides little benefit when N is not being mineralized.

There are other lines of evidence that support the dominant role of transitive limitation at Konza. For example, regardless of whether you add N or water, the same species--Panicum virgatum--comes to dominate. If N was limiting, wouldn't adding N dry out the soils more and favor a low-water, high-N species?

One of the tough things to demonstrate is the roll that soil water potential plays in productivity. I'll likely expand on this later, but there are no relationships between water potential and productivity, only conductance. If we could show that productivity should not be diminished by lowering soil water potential to say -2 MPa, we might be able to demonstrate that it is not water that is limiting directly, but transitively by reducing N supplies.

There are still multiple pieces to assemble before the story is complete, but transitive limitation is likely a linchpin in understanding grasslands.

Transitive limitation and precipitation

Thinking more about N and water, I was looking over the Huxman-Smith et al. 2004 Nature paper. This paper summarizes the sensitivity of ANPP to precipitation. With data from 14 sites, they calculate rainfall use efficiencies across years to see how RUE changes with mean annual precipitation. They find that wetter sites have lower RUE, but all sites converge on a constant RUE. (Above figure is precipitation vs. ANPP (g m-2), each x-axis tick is 500 mm y-1).

The most interesting art of the paper are statements on resource limitation.

Here are some key sentences:

1) the authors predict that “the removal of other resource limitations so that precipitation becomes the primary limiting resource will result in an increase in site-level RUE that approaches RUEmax.”

2) “sites with high production potential in years with greater than average precipitation, soil nitrogen or other limiting resources might transiently limit biological activity.”

3) They also state that “biogeochemical constraints (limitation of activity by resources other than water) can increase with increasing precipitation”.

The approach the authors take to limitation is generally one of serial limitation. First one resource limits, and then another. The authors seem to hold the idea that multiple resources can co-limit ANPP, but they are mute about mechanisms or what the tradeoffs are. There is no evidence they considered substitutability leading to co-limitation, or transitive limitation (water limiting N availability). Looking at statement 3 above, they just as easily could have said that biogeochemical constraints can also decrease with increasing precipitation.

The unstated model they use follows something like this: less rain, greater water stress, less limitation by other resources, less production, greater WUE.

The existence of transitive limitation changes the entire story of ANPP responses to increased precipitation: less rain, less nitrogen mineralized, greater limitation by N, less production.

The relationship between ANPP and precipitation is, here, just a ratio. The important part of research is always to look one level of mechanism below the pattern of interest.

For example, if plant WUE was a key factor in RUE, then wouldn’t sites dominated by C4 vegetation have an inherently higher RUE than sites dominated by C3 vegetation? But, if the observed RUE was driven by N mineralization responses to increased soil moisture, wouldn’t they be the same? Although Cedar Creek (CDR) might show no increase in ANPP with increased precipitation due to strong N limitation, there might be unique patterns of sensitivity of N mineralization to variation in precipitation there. For example, maybe with the sandy soils, soil moisture isn't much greater in a high precipitation year.

At this point, both mechanistic hypotheses should be considered equally (one can’t be favored because it comes first). It’d be interesting to see how much of the patterns could be explained by transitive limitation.

Huxman, T. E., M. D. Smith, P. A. Fay, A. K. Knapp, M. R. Shaw, M. E. Loik, S. D. Smith, D. T. Tissue, J. C. Zak, J. F. Weltzin, W. T. Pockman, O. E. Sala, B. M. Haddad, J. Harte, G. W. Koch, S. Schwinning, E. E. Small, and D. G. Williams. 2004. Convergence across biomes to a common rain-use efficiency. Nature 429:651-654.