Precipitation variability and primary productivity in water-limited ecosystems: how plants ‘leverage’ precipitation to ‘finance’ growth


In strongly water-limited ecosystems, water to support primary productivity and other aspects of ecosystem function is, by definition, in short supply. It is widely recognized that water limitation partly reflects low total quantities of rainfall on an annual or a growing season basis. However, annual precipitation quantities represent one level in a hierarchy of temporal scales of precipitation variability (Greenland, 1999). It is becoming increasingly recognized that variation in the seasonality of precipitation, the timing between precipitation events and the quantity of rainfall per event can be as important as annual quantity for net primary productivity and other ecosystem processes (Knapp et al., 2002, 2008). This topic is particularly important in light of continuing climate change. Increases in both total rainfall amounts and in the frequency of extreme events have been documented and are likely to continue during the 21st century (Groisman et al., 2005, Christensen & Hewitson, 2007, Groisman & Knight, 2008). Studies on primary productivity/precipitation relationships often focus on total above-ground net primary productivity (ANPP) because of its coupling with biogeochemical cycles and with the atmosphere. In this issue of New Phytologist, Robertson et al. (pp. 230–242) make an important contribution by going beyond total ANPP to examine the effects of interannual and seasonal variation in precipitation inputs on the productivity of three dominant plant species in a desert grassland in the southwestern USA. Their results reveal the complexity of species responses to precipitation variability, reflecting both feedback from precipitation variability on other growth-limiting resources and the life-history adaptations of these species.

‘... to understand how precipitation patterns influences the water limitation of ecosystem productivity one needs to understand the responses at the species and functional group levels.’

Robertson et al. studied a sotol grassland ecosystem in the Chihuahuan desert in southwestern Texas, USA. Their study site receives approx. 365 mm of mean annual rainfall, mostly in the summer and in the fall. They chose three dominant perennial plant species for study: a shrub, Dasylirion leiophyllum (‘sotol’); a succulent, Opuntia phaeacantha (‘brown spine prickly pear’); and a grass, Bouteloua curtipendula (‘sideoats grama’). Several traits make these species likely to differ in response to variation in seasonality and annual amount of precipitation. The shrub Dasylirion has an expansive root system (with both dense fibrous roots at the surface and well-developed deep roots), a woody caudex that can store water and reaches peak biomass during summer. The succulent, Opuntia, has a shallow root system (mostly 10–30 cm deep), stores water in its fleshy pads and reaches peak biomass in spring to early summer. The grass, Bouteloua, is a shallow, fibrous-rooted bunchgrass reaching peak biomass in mid to late summer.

Plots containing these species were subjected to precipitation treatments representing expected future precipitation patterns for the region: natural precipitation plus either supplemental winter precipitation or supplemental summer precipitation, or both. Additional control plots received only natural precipitation. These treatments were applied for 5 yr, which is itself no mean feat given the ruggedness and isolation of these ecosystems. Above-ground net primary productivity for each species was estimated several times each year, as were as soil inline image and inline image, and several other parameters.

Desert grasslands would be expected to be more sensitive to interannual variation in precipitation compared with more mesic grasslands or forests (Fig. 1; Huxman et al., 2004). Indeed, for all these species, Robertson et al. found statistically significant (P ≤ 0.10) differences among years in ANPP. But surprisingly, there was no correlation between ANPP and annual precipitation for Dasylirion or Opuntia, and ANPP actually decreased with increasing annual precipitation for Bouteloua. As these results imply, the water-supplementation treatments also had no effects on species ANPP except in Opuntia, where summer + winter supplementation increased ANPP over that of control plots.

Figure 1.

Sensitivity of total above-ground net primary productivity (ANPP) to interannual variation in total annual precipitation from 14 sites with long-term ANPP records. The overall regression of ANPP vs precipitation for all sites and years combined was highly significant (P < 0.001). Individual sites (noted by three-letter identifiers) varied in their sensitivity to between-year variation in precipitation, with desert sites showing much higher sensitivity to precipitation than those receiving greater annual precipitation totals (inset). Site abbreviations and other details in Huxman et al. 2004.

Feedbacks from precipitation on resource availability

Control of ANPP for these species was not consistently the annual quantity or seasonality of precipitation inputs. This result seemed to be counterintuitive because seasonal and interannual variability is often viewed as the most relevant temporal scale of variability for net primary productivity and species changes (Schwinning & Sala, 2004). Precipitation variability also has feedbacks on other plant resources, especially nitrogen (N) availability. The microbial processes regulating nutrient availability are especially sensitive to short-term variation in soil moisture as a result of varying event patterns (Schwinning & Sala, 2004). This is because, as soils dry and rewet between rain events, the balance between mineralization and immobilization of N is altered, and soils undergoing cyclical wet–dry periods often have higher N-mineralization rates than continuously moist soils (Austin et al., 2004). Thus, precipitation variability not only affects the size of the pool of water in the soil, it also affects the size of the available N pool. Further complicating matters, these two pools do not vary in synch, sometimes peaking together, sometimes peaking apart (Seastedt & Knapp, 1993), creating nonlinear dynamics in temporal trends in ANPP.

Robertson et al. next analyzed how precipitation pattern (number of events, event size and event interval), temperature and soil variables may have influenced each species’ ANPP in wet, average and dry years. They found some important feedbacks among precipitation variability, N availability and species ANPP. For example, control of Dasylirion ANPP appeared to oscillate between N availability in wet years and water availability in dry years, with one becoming progressively limiting as the other increased; essentially the Progressive Nitrogen Limitation hypothesis (Luo et al., 2004) applied to water and N instead of to CO2 and N.

In contrast, for Bouteloua, the analysis suggested feedbacks among precipitation amount, pattern and N availability. In wetter-than-average years, the ANPP increased with higher soil N and longer times between precipitation events. This may reflect enhanced N mineralization from greater soil moisture variability (Austin et al., 2004). Thus, Bouteloua also appeared to exhibit temporally shifting controls on productivity, but in this case, the feedbacks were between precipitation pattern and N, rather than between precipitation amount and N, as for Dasylirion.

For Opuntia, the analysis suggested that ANPP was primarily regulated by precipitation seasonality and pattern rather than by N availability or other soil attributes. For example, in dry years and in years with average precipitation, the ANPP of Opuntia increased with small, more frequent, precipitation events and winter supplementation, and decreased in some years when long interpulse periods probably led to drying of Opuntia's main rooting zone.

Resource feedbacks and plant adaptations to water-limited environments: leveraging scarce capital to support growth

Clearly then, the role of precipitation amounts and patterns and feedbacks with N availability differed for these three species, and the ability to acquire moisture differed according to each species’ life-history attributes. To illustrate, consider an analogy from current world economic affairs, where a lack of available credit threatens to lower human economic productivity (recalling that both ‘economy’ and ‘ecology’ derive from the same root, the Greek oikos, ‘house’). We can heuristically consider the precipitation captured in the soil as ‘capital’, and the plants as borrowing that capital, for it is eventually returned. Plants thus become ‘leveraged’ to ‘finance’ growth. As with the current financial situation, plant ‘capital’ can become scarce, sometimes with little warning! The question, from the plant's perspective, is how to deal with imperfectly matched ‘capital’ availability, relative to when the plant most needs it. Just like businesses, plants cope with this mismatch in different ways, with some having many sources of credit to draw on, some able to maintain large capital reserves and some living with chronic cash-flow problems.

The three species in the study of Robertson et al. represent these three ways of coping with temporal variability in the availability of water. Dasylirion exemplified the ‘many sources of credit’ approach. Its combination of both deep and shallow roots made Dasylirion ANPP more dependent on overall levels of resource availability than on variation in precipitation pattern. The succulent, Opuntia, fits the second approach. Opuntia's shallow root system made it more subject to ‘capital’ scarcity and therefore its growth was sensitive to the seasonality and pattern of precipitation. However, water storage in its fleshy pads made Opuntia relatively free of the influences of interannual variation in precipitation and variation in N availability. Bouteloua, then, suffers from cash-flow problems. A shallow root system and lack of capacity to store any ‘capital’ reserves made it live close to the edge in terms of access to moisture. Like a business with chronic cash-flow problems, Bouteloua may fail to produce at all if ‘capital’ becomes chronically unavailable.

The most important message of the article by Robertson et al. is that to understand how precipitation patterns influences the water limitation of ecosystem productivity, one needs to understand the responses at the species and functional group levels. This will help to reveal which precipitation patterns may promote species stability and which may lead to species replacement or even cause community turnover to a fundamentally different assemblage. It is also crucial that future research addresses how temperature, CO2, or edaphic factors may modify plant responses to precipitation variability. As precipitation is coupled with the processes controlling N availability, ecosystem responses to climate change will depend on how the strength of that coupling varies under future precipitation regimes.