Elevated CO2 effects
Overall CO2 effects on fine root standing crop, production, and loss were not significant in our study. Thus, over long periods, we reject our first and third hypotheses: elevated CO2 did not increase fine roots for either the two dominant Mojave desert shrubs (A. dumosa, L. tridentata) or for the plant community. These results in an arid ecosystem differ from those of most FACE experiments in agricultural and mesic ecosystems. Elevated CO2 typically increased root biomass for agricultural species grown under FACE experiments, with average root biomass increases of 25%, 47% and 64% for clover, grass crops and cotton, respectively (Kimball et al., 2002). Increases in fine roots often were observed in forest FACE experiments (Norby et al., 2004; Iversen et al., 2008; Pregitzer et al., 2008; Pritchard et al., 2008; Jackson et al., 2009; Liberloo et al., 2009) and in a grazed pasture FACE experiment (Allard et al., 2005). Many open-top chamber studies in grasslands also show greater fine roots under elevated CO2 (Pendall et al., 2004; Sindhøj et al., 2004; Milchunas et al., 2005; LeCain et al., 2006; Volder et al., 2007; Anderson et al., 2010). However, increased fine roots under elevated CO2 were not uniform or did not occur in some FACE studies. For California annual grasslands, increased fine roots under elevated CO2 only occurred at the end of the growing season, presumably owing to increased soil water under elevated CO2 (Higgins et al., 2002). For FACE experiments in mature deciduous forests (Bader et al., 2009) or at treeline (Handa et al., 2008), elevated CO2 did not affect fine roots or even suppressed them.
Although the overall CO2 effect was not significant, significant CO2 effects on root standing crop occurred on individual sampling dates during winter and spring of 2004–05 (Fig. 2). However, these CO2 treatment differences in standing crop need to be interpreted cautiously. Fine roots persist for long periods in desert soils (Table 5), and the extent that fine roots maintain physiological function over a several-year period is unknown, but unlikely. Thus, greater fine root standing crop does not necessarily indicate greater access to soil resources. Furthermore, differences in standing crop were transitory and were related more to differences in fine root loss than production (Fig. 3). For example, significantly greater standing crop under elevated CO2 for Ambrosia in December 2004 (Fig. 2a) occurred largely because of less root loss under elevated CO2 during July–December 2004 (Fig. 3a), and differences in standing crop became nonsignificant during spring 2005 largely because root loss under elevated CO2 was significantly greater than ambient during January–June 2005. Thus, these transient CO2 effects are not consistent with the mechanism underlying the first and third hypotheses, that is, greater fine root production under elevated CO2.
We also reject the second hypothesis that fine root persistence increases and root turnover decreases under elevated CO2. Fine root persistence was similar between CO2 treatments for all microsites (Table 5), and the probability of root loss was not significantly affected by CO2 treatment (Table 6). Although annual root turnover was significantly lower under elevated CO2 for Ambrosia and Larrea in 1 yr (Table 4), root turnover under ambient CO2 (0.38 ± 0.06) was nearly identical to that under elevated CO2 (0.39 ± 0.07) when averaged over all sampling locations and all 4.5 yr of measurements. These results from the Mojave desert differ from those in some, but not all, elevated CO2 studies. For example, decreased fine root life spans under elevated CO2 were reported in both grasslands (Allard et al., 2005; Anderson et al., 2010) and forests (Pritchard et al., 2008). Increased annual root turnover under elevated CO2 has been reported (Liberloo et al., 2009), but turnover was not significantly affected in other ecosystems (Higgins et al., 2002).
Lastly, in contrast to forested ecosystems (Iversen, 2010), we also reject the fourth hypothesis that elevated CO2 shifts root distribution to deeper within the soil profile in the Mojave desert. The only significant CO2 treatment effect on fine root production below 0.25 m soil depth occurred for the 0.75–1.0 m depth interval during January–June 2005 (Figs 4–6), but increased root production occurred under ambient, rather than elevated, CO2 at all sampling locations. Our results also indicate general lack of fine root production below 0.75 m soil depth, which may seem surprising given the known ability of Ambrosia and Larrea to produce roots to 5 m soil depth (Hartle et al., 2006) as well as greater maximum rooting depths generally found in deserts (Jackson et al., 1996; Jobbagy & Jackson, 2000). However, January–June 2005 was the only period during the 4.5 yr of study that soil water was consistently available below 0.65 m soil depth (Fig. S1). Because percolation of precipitation to deep within the soil profile is rare in the Mojave desert, we suspect that potential shifts to deeper fine roots under elevated CO2 is at least as rare. Furthermore, the propensity of desert ecosystems to extract all available water from the soil profile (Anderson et al., 1987, 1993; Yoder & Nowak, 1999), even under elevated CO2 (Nowak et al., 2004b), may ultimately constrain fine root production.
Two unanticipated effects of elevated CO2 on Mojave desert vegetation, earlier peaks in fine root production (Table 3) and greater amount of fine root production in shallow soils for the plant community during an above-average precipitation year (Fig. 6), appear to be ecologically important. Growth of both annuals (Smith et al., 2000) and perennials (Smith et al., 2000; Housman et al., 2006) are accelerated under elevated CO2, especially in wetter years. Increased C assimilation in these wetter years (Huxman et al., 1998; Hamerlynck et al., 2000; Naumburg et al., 2003; Housman et al., 2006), coupled with earlier fine root growth (Table 3), help provide C, water and nutrient resources needed to sustain earlier growth. Lower water content of shallow soils under elevated CO2 during spring 2005 (Fig. S1) is consistent with this concept that earlier shoot growth is sustained by greater extraction of water from soil by greater amounts of fine roots earlier in the growing season.
The CO2 effects on fine root dynamics in our current study from 2003 to 2007 were similar in most regards to those reported by Phillips et al. (2006) for 1998–99 at our site. For Ambrosia and Larrea, Phillips et al. (2006) found no significant overall CO2 treatment effects nor any significant CO2 × time interactions on fine root standing crop, production, and loss. Our results (Tables 1, 2) also show no overall CO2 effect, although greater standing crop under elevated CO2 occurred for Ambrosia for a short period in early 2005 (Fig. 2) and production and loss for both species were occasionally significantly different between CO2 treatments (Fig. 3). For the plant community as a whole, fine root standing crop under elevated CO2 was significantly less than ambient CO2 during 1998–99 (Phillips et al., 2006), but was significantly greater for c. 6 months in late 2004 and early 2005 in our study (Fig. 2). For annual root turnover, Phillips et al. (2006) found no significant differences between CO2 treatments for Ambrosia and Larrea but significantly lower under elevated CO2 for the community. In our study, significant CO2 treatment effects occurred in one of 4 yr of measurement for Ambrosia and Larrea, but no significant differences occurred for the community. Increased root biomass has been reported for Larrea seedlings grown in controlled environments under elevated CO2 (Obrist & Arnone, 2003; Clark et al., 2010), but positive effects of elevated CO2 decreased through time (Obrist & Arnone, 2003). Responses of three Mojave desert grasses to elevated CO2 also were examined in controlled environment experiments (Yoder et al., 2000): only a C3 perennial tussock grass had significant increase in root biomass under elevated CO2, whereas the root biomass of a C3 invasive annual grass and of a C4 perennial tussock grass were not significantly affected by elevated CO2. Thus, taken as a whole, results indicate that significant effects of elevated CO2 on root system dynamics can occur in some Mojave desert species, but CO2 effects are rare and transient.
Fine root dynamics
As in many other ecosystems, fine root production in the Mojave desert was strongly seasonal, peaking between February and May with peak production both greater and earlier in wet years. Similar to shortgrass steppe ecosystems (Milchunas et al., 2005), fine root loss was not as strongly influenced by time of year or precipitation as production, and survival and proportional hazard analyses indicate that roots typically are slowly lost through time. By contrast, mesic forest (Norby et al., 2004) and grassland (Sindhøj et al., 2004) ecosystems have seasonal losses of fine roots that start near the end of the growing season.
Very long persistence of fine roots in the Mojave desert also differs from those in other ecosystems. In our study, 50% of fine roots persisted for two or more years, and 7% of fine roots persisted for > 4 yr. In comparison, the longest time to 50% mortality that was reported in a review of annual crops, perennial herbaceous crops, woody fruit crops and forest communities was just short of 1 yr, with results from 22 of 34 studies indicating 50% mortality in < 100 d (Eissenstat & Yanai, 1997). In more recent publications, 50% of fine roots survived only 2–3 months for deciduous (Black et al., 1998; Wells & Eissenstat, 2001) and evergreen (Black et al., 1998; Johnson et al., 2000) trees and for grapes (Anderson et al., 2003). For three grass and shrub species in a shrub steppe ecosystem, 50% of fine roots typically survived for 3–6 months (Peek et al., 2005). However, not all trees have fine roots with short life spans (Eissenstat et al., 2000; Pritchard et al., 2008; Strand et al.; 2008), and even grasslands (Milchunas et al., 2005) have fine roots whose life spans approach 2 yr. Many factors affect fine root longevity, including root (root diameter, root order, etc.) and environmental (temperature, soil water content, etc.) characteristics (Eissenstat & Yanai, 1997; Black et al., 1998; Johnson et al., 2000; Wells & Eissenstat, 2001; Anderson et al., 2003; Peek et al., 2005). Long persistence of roots in the Mojave desert does not appear to be related to root diameter or root order characteristics: root longevity typically decreases with decreased root diameter and order, but root diameters in our study were as small as or smaller than those in most other studies and most fine roots in our study appeared to be first-order roots. Because fine root persistence through winter can be very low (Wells & Eissenstat, 2001), the lack of severe winters with cold air temperatures and freezing soils in the Mojave desert may partially account for long persistence of fine roots in our study. We also suspect that slow decomposition rates in extremely dry desert soils delays loss of fine roots to the soil organic matter pool. Thus, fine roots persist for long periods in the Mojave desert even though fine roots are not physiologically functional for the entire time they are present.
Production and loss of fine roots in shrub sampling locations typically was similar to that in the community as a whole except for spring 2005 when Larrea produced and lost greater amounts of fine roots. Phillips et al. (2006) also observed similar root production and loss among the three sampling locations. Given that > 80% of the ground surface is bare soil at the NDFF (Jordan et al., 1999) and that the majority of the community transect minirhizotron tubes were located in areas with bare soil, it may seem surprising that fine root production in areas without perennial plants (and often few annual plants) that is sampled by the community transects would have similar amounts of fine roots as that under perennial, dominant shrubs. However, shrubs in the Mojave desert have horizontal rooting that typically exceed 1 m and can extend beyond 5 m (Hartle et al., 2006), and fine root growth of some shrubs can be as great or greater in the interspace as under the shrub canopy (Wilcox et al., 2004). Thus, although much of the desert surface is bare soil interspace between plants, roots of surrounding plants readily explore and extract resources from these unvegetated interspaces.
Our results differ from earlier studies at the NDFF in that our initial fine root standing crop (January 2003) was higher than the final standing crop in January 2000 reported by Phillips et al. (2006). Phillips et al. (2006) reported a net increase of c. 25 m m−2 in fine root standing crop for all sampling locations over 2 yr. Given the 3-yr gap between the end of the Phillips et al. study and the beginning of our study coupled with the numeric (but not significant) difference in fine roots observed on the top of tubes (Phillips et al., 2006) and on the side of tubes (this study) (see Supporting Information, Methods S1 for details), our initial standing crops of c. 75 m m−2 would be consistent with a gradual accumulation of fine roots through time, as observed by Phillips et al. and by us during the first 6 months of our study (Figs 2, 3). Largely because standing crop was greater in our study than in Phillips et al. (2006), our estimates of annual root turnover are much smaller than theirs: their smallest mean annual root turnover (2.33) is over three times greater than the greatest mean root turnover in our study (0.75). Because absolute production and loss rates are similar between the two studies, annual production and turnover indices must be interpreted with caution in ecosystems where fine roots persist for periods of time > 1 yr. Furthermore, although Johnson et al. (2001) suggest that it may take 6–12 months for fine roots to recover from tube installation, equilibration may take significantly longer, as suggested by Strand et al. (2008).
Implications for carbon cycling
Carbon dynamics of ecosystems are influenced by root loss (Eissenstat et al., 2000; Strand et al., 2008): root decomposition increases C input to soil organic matter. In our study, fine root loss reached maximum values for both CO2 treatments in late 2004 and early 2005 (Fig. 3). These large losses of roots, which indicates movement from the fine root pool to the soil organic matter pool, also corresponds with a period of above-average precipitation (Table 3). Soils at the end of summer in 2004 were very dry (Fig. S1), but a series of storms beginning in mid-October 2004 and continuing into May 2005 kept soils relatively wet from late 2004 to spring 2005, providing warm and wet soil conditions ideal for microbial transformation of fine roots into soil organic matter. Once initiated, transformation of fine roots into soil organic matter continued over the next 2 yr even though soils were relatively dry during much of this period. Thus, the rate of soil C input from decomposition of fine roots in the Mojave desert was temporally variable. However, total C input from fine root litter over long periods should be similar between ambient and elevated CO2 treatments: minirhizotron measurements of root length per area of tube sidewall (Phillips et al., 2006; this study) as well as measurements of specific root length and fine root C (Clark et al., 2010) were not significantly affected by elevated CO2. Thus, differences in nutrient cycling (Jin & Evans, 2007) and microbial communities (Jin & Evans, 2010) between ambient and elevated CO2 at our Mojave desert study area indicate that soil microbial factors may be more important determinants of C balance below ground than input of fine root litter.