Direct and legacy effects of long-term elevated CO2 on fine root growth and plant–insect interactions



  • Increasing atmospheric CO2 concentrations alter leaf physiology, with effects that cascade to communities and ecosystems. Yet, responses over cycles of disturbance and recovery are not well known, because most experiments span limited ecological time. We examined the effects of CO2 on root growth, herbivory and arthropod biodiversity in a woodland from 1996 to 2006, and the legacy of CO2 enrichment on these processes during the year after the CO2 treatment ceased.
  • We used minirhizotrons to study root growth, leaf censuses to study herbivory and pitfall traps to determine the effects of elevated CO2 on arthropod biodiversity.
  • Elevated CO2 increased fine root biomass, but decreased foliar nitrogen and herbivory on all plant species. Insect biodiversity was unchanged in elevated CO2. Legacy effects of elevated CO2 disappeared quickly as fine root growth, foliar nitrogen and herbivory levels recovered in the next growing season following the cessation of elevated CO2.
  • Although the effects of elevated CO2 cascade through plants to herbivores, they do not reach other trophic levels, and biodiversity remains unchanged. The legacy of 10 yr of elevated CO2 on plant–herbivore interactions in this system appear to be minimal, indicating that the effects of elevated CO2 may not accumulate over cycles of disturbance and recovery.


Global atmospheric carbon dioxide (CO2) levels continue to increase rapidly, mainly because of the burning of fossil fuels. The atmospheric concentration of CO2 has increased from a preindustrial level (c. 1750) of c. 270 ppm to a current level of c. 394 ppm, an increase of 124 ppm, or 45%. Most studies indicate that CO2 levels will at least double from preindustrial levels over the next five to ten decades. This increase represents one of the most large-scale and wide-reaching perturbations to the environment (IPCC, 2007).

Many studies have shown an increase in above- and below-ground plant growth in elevated relative to ambient CO2 (Curtis & Wang, 1998; Norby et al., 1999; Long et al., 2004; Ainsworth & Long, 2005; Jackson et al., 2009; Seiler et al., 2009; Day et al., 2013). However, elevated CO2 inhibits the assimilation of nitrate into organic nitrogen compounds (Bloom et al., 2010) and usually reduces plant nitrogen concentrations and increases secondary metabolites (Lincoln et al., 1993; Poorter et al., 1997; Curtis & Wang, 1998; Bidart-Bouzat & Imeh-Nathaniel, 2008; Zavala et al., 2013). This decreased plant quality decreases herbivore performance and levels of herbivory in many communities (Stiling & Cornelissen, 2007; Lindroth, 2010; Robinson et al., 2012). Theoretically, the effects of elevated CO2 on insect herbivores may cascade up to impact species feeding at higher trophic levels, such as spiders and parasitic wasps. In addition, increased biomass, and therefore litter production, could provide increased resources for detritivores. These effects could lead to changes in arthropod biodiversity.

Most studies investigating the effects of elevated CO2 on plants and plant–herbivore interactions have used short-term experiments comparing the digestion and performance of insect herbivores fed foliage grown in elevated or ambient CO2 (Barbehenn et al., 2004; Sudderth et al., 2005; Agrell et al., 2006). Fewer studies have used elevated CO2 levels in the field and measured changes in plant growth, herbivory and biodiversity, despite the greater validity of this technique to natural conditions (Sanders et al., 2004; Hillstrom & Lindroth, 2008; Stiling et al., 2010). Finally, no studies have investigated the legacy effects of elevated CO2 on plant growth and plant–insect interactions. What happens to plant growth, plant quality and herbivory when elevated CO2 levels are terminated? Extrapolations from previous work suggest that legacy effects of elevated CO2 could last for a considerable period of time. For example, a substantial below-ground carbon sink develops that can affect plant re-growth for many years (Canadell et al., 1996; Lousteau et al., 2001). In addition, increased microbial turnover in elevated CO2, together with nitrogen accumulation in above-ground plant biomass, can cause progressive nutrient limitation over long time periods, depressing plant growth (Gifford et al., 1996; Cannell & Thornley, 1998; Hungate et al., 2006). Thus, it is possible that long-term legacy effects of elevated CO2 could affect plant growth and plant–insect interactions for many years after the cessation of elevated CO2.

This article provides data on the effects of elevated CO2 on plants and insect herbivory in a Florida scrub oak forest after many years of continuously elevated CO2. In addition, we provide data on the richness of insect orders and of beetle families and functional groups to examine whether there are noticeable changes in biodiversity under elevated CO2. Finally, we examine the legacy effects of elevated CO2 on plant re-growth and insect herbivory by measuring fine root re-growth and herbivory levels after elevated CO2 is discontinued.

Materials and Methods

This study was conducted at NASA's Kennedy Space Center, in a scrub-oak, palmetto ecosystem (Schmalzer & Hinkle, 1992). The study site was dominated by three species of oak: myrtle oak, Quercus myrtifolia Willd, sand-live oak, Q. geminata Small, and Chapman oak, Q. chapmanii Sargenti, which together accounted for 85–90% of the plant biomass (Seiler et al., 2009). Of the three species of oak, Q. myrtifolia was the most common, accounting for c. 77% of the oak biomass, whereas Q. geminata accounted for c. 17% and Q. chapmanii for 6% (Dijkstra et al., 2002). These oaks rely on stored below-ground resources for many years during stand development (Langley et al., 2002). Twenty-seven additional plant species were identified in the area with Elliott's milk pea, Galactia elliottii, Nuthall, a nitrogen-fixing legume, the most common. Scrub-oak is a fire-dependent system and, at Kennedy Space Center, the community exists on a fire return cycle of between 10 and 15 yr.

In January 1996, a 0.8-ha area of scrub-oak forest was burned. During the following spring, 16 open top chambers (OTCs), each 2.5 m in height with octagonal sides of 1.4 m in width, were erected in the study area. Each side consisted of a 10-cm PVC pipe frame covered with Mylar (Melinex 071; Courtaulds Performance Films, Martinsville, VA, USA). Panels were easily removable to facilitate entry into the chambers. Eight OTCs were maintained at elevated CO2 (ambient + 350 ppm CO2) and eight at ambient CO2 (c. 350 ppm in 1996 to c. 380 ppm in 2007). Ambient or CO2-enriched air was blown into each chamber via four 20.5-cm-diameter ducts at a rate of 24–30 m3 min−1. The blower speed was reduced at night to one-third of the daytime values. These CO2 concentrations were continuously maintained inside the chambers from mid-May 1996 through mid-June 2007 (except for small periods in 1999 and 2005 when the chambers were damaged by hurricanes). The effects of elevated CO2 on above- and below-ground plant growth, and on nutrient cycling, are presented elsewhere (Seiler et al., 2009; Day et al., 2013). Here, we present previously unpublished data on the effects of elevated CO2 on plant–herbivore interactions, insect biodiversity and the legacy effects of elevated CO2 on fine root growth and herbivory.


During August 2001 and 2002, we counted the numbers of leaf miners, leaf tiers, chewed leaves, eyespot galls, leaf galls and leaves with pathogens per 200 haphazardly selected leaves on each oak species in each chamber and on the legume Galactia elliottii. Among the most common leaf mining genera were Cameraria, Stigmella, Stilbosis and Buccalatrix. Leaf damage was caused by various chewing lepidopteran and orthopteran species, leaf tying by various lepidopterans, eye spot galls by cecidomyiids and other leaf galls by cecidomyiids, including Belonocnema quercusvirens, Neuroterus quercusverrucarum, Sphaeroteras carolina, Smelleum and others. Leaf pathogens were not identified.

Statistical analyses of the effects of CO2 treatment on damaged leaves for 2001 and 2002 were performed using split-plot ANOVAs on the total numbers of leaves damaged by each herbivore guild or pathogen, with CO2 as the main factor, guild and plant species as the subplot factors and chamber as a random effect.


Pitfall trap catches are considered to be good indicators of biodiversity in most terrestrial habitats (Duelli et al., 1999; Hillstrom & Lindroth, 2008). We installed two 8.5-cm-diameter × 6-cm-deep pitfall traps per chamber. Traps were half filled with antifreeze to keep insects from crawling out and to minimize fluid loss through evaporation. Traps were installed at the end of 2002 and were replaced approximately bimonthly for 3 yr, until 2005. All arthropods were identified to order under a dissection scope. In addition, in 2004 samples, all beetles were identified to family. The biodiversity of trap catches was analyzed using repeated-measures ANOVA of bimonthly totals of arthropod orders or beetle families. In addition, in 2004, we scored beetles as herbivores, detritivores, insectivores and fungivores and analyzed treatment effects using Wilk's lambda MANOVA.

Legacy effects

The chambers were dismantled and all vegetation was harvested in July 2007 to determine species-specific and community biomass responses to 11 yr of elevated CO2 (Seiler et al., 2009). During the remainder of 2007, and in 2008, the vegetation began to re-grow from the remaining roots under ambient atmospheric CO2 levels.

To estimate the legacy effects of elevated CO2 on fine root growth, images from minirhizotrons installed in the former chamber plots were collected in August 2007 (c. 1 month after above-ground vegetation removal) and May 2008 (c. 10 months after removal) using the methods described in Day et al. (2013). Digital jpeg images were captured from the video recordings. Fine root biomass (g m−2 to a depth of 1 m) was calculated from root length and width values for all roots < 2 mm in diameter, following the methods detailed by Day et al. (2013). For statistical analyses, the data were log-transformed to meet the assumptions for ANOVA. Fine root biomass was tested with a four-factor repeated-measures ANOVA using SAS Proc GLM (SAS version 9.1; SAS Institute Inc., Cary, NC, USA), with plot as the random effect and CO2 treatment, depth and date as fixed effects. A three-factor nested ANOVA was run on each individual date to test for CO2 treatment effects; plot was the random effect and treatment and depth were fixed effects.

During September 2008, we counted the numbers of leaf mines and chewed leaves per 200 haphazardly selected leaves on Q. myrtifolia, Q. chapmanii, Q. geminata and G. elliottii in each ambient or elevated CO2 legacy plot. In addition, leaves of each species were collected haphazardly throughout the plots and oven dried at 70°C, and then ground and analyzed for percentage nitrogen. Statistical analyses of the legacy effects of CO2 treatment on the numbers of leaf mines per 200 leaves, number of chewed leaves per 200 leaves and percentage leaf nitrogen were performed using split-plot ANOVAs with CO2 as the main plot factor and chamber as a random effect. Three chambers were omitted from the percentage leaf nitrogen analyses because no Galactia was collected.



There was a significant effect of CO2 on leaf damage: elevated CO2 reduced the numbers of leaves damaged by leaf miners, leaf tiers, leaf chewers, eyespot galls and other leaf galls for all four plant species in both 2001 (Fig. 1, P < 0.001) and 2002 (Fig. 2, P < 0.001). There was also a significant effect of tree species in both years, as the amount of herbivore damage varied between host plant species (P < 0.001 for both years), but there was no interaction of CO2 with plant species, meaning that elevated CO2 depressed leaf damage on all plant species (2001, = 0.793; 2002, = 0.808). There was also a significant effect of guild on leaf damage, because damage by some guilds, such as leaf miners and leaf tiers, was more common than by others (Figs 1, 2, P < 0.001 for both years). There was an interaction of CO2 with guild (2001, = 0.002; 2002, = 0.030), as pathogen-damaged leaves were not consistently depressed in elevated CO2, but all other types of insect-damaged leaves were. Finally, there was a significant interaction of guild and tree species (2001 and 2002, both < 0.001), as the abundance of leaves damaged by different guilds varied according to tree species, but there was no three-way interaction between CO2 level, guild and tree species (2001, = 0.966; 2002, = 0.924).

Figure 1.

Frequency of damage per 200 leaves, ± SEM, September 2001, in ambient (open bars) and elevated (closed bars) CO2 at Kennedy Space Center, FL, USA. (a) Quercus myrtifolia; (b) Q. geminata; (c) Q. chapmanii; (d) Galactia elliottii.

Figure 2.

Frequency of damage per 200 leaves, ± SEM, September 2002, in ambient (open bars) and elevated (closed bars) CO2 at Kennedy Space Center, FL, USA. (a) Quercus myrtifolia; (b) Q. geminate; (c) Q. chapmanii; (d) Galactia elliottii.


Arthropods from 25 orders were found in pitfall traps, but there was no significant effect of CO2 treatment on arthropod order richness (= 1.000), although richness varied through time (Fig. 3a, P < 0.001). Beetles from 39 families were found in pitfall traps. There was also no significant effect of elevated CO2 on beetle family richness in pitfall traps in 2004 (= 1.000), although beetle richness also varied over time (Fig. 3b, P < 0.001). For beetles, there were no significant effects of elevated CO2 on any guild or interaction of time with CO2 (Fig. 4, P > 0.05 in all cases).

Figure 3.

Arthropod richness in pitfall traps, ± SEM, under ambient (open circles) and elevated (closed circles) CO2 at Kennedy Space Center, FL, USA for (a) arthropod orders and (b) beetle families.

Figure 4.

Beetle densities per chamber, ± SEM, during 2004 for herbivores, detritivores, insectivores and frugivores. Open circles, ambient CO2; closed circles, elevated CO2.

Legacy effects

Fine root biomass values in ambient CO2 plots were 1644, 1620 and 1687 g m−2 for March 2007, August 2007 and May 2008, respectively. In elevated CO2 plots, fine root biomass values were 1942, 1852 and 2078 g m−2 for the same time series (Fig. 5). No statistically significant CO2 treatment effect on fine root biomass was detected on any given date (March 2007, = 0.31; August 2007, = 0.57; May 2008, = 0.39), although fine root biomass was consistently higher in plots previously under elevated CO2 for all three sample dates. However, there was a significant difference among the three sampling dates (< 0.0001) in the previously elevated CO2 plots. There was minimal change in fine root biomass in the ambient plots over the three sample dates. Fine root biomass increased by only 4% in the ambient plots between August 2007 and May 2008, but increased by 12% in the formerly elevated CO2 plots, indicating significant recovery of fine root biomass in the elevated plots, but not in the ambient plots.

Figure 5.

Fine root biomass (g m−2) to 100-cm depth estimated using minirhizotrons for sampling dates before and after complete above-ground vegetation removal. Values are means + SE. Open bars, ambient CO2; closed bars, elevated CO2.

Leaf nitrogen was unaffected by previous CO2 treatment (Fig. 6, P = 0.760) and, although there was an effect of plant species on foliar nitrogen (< 0.001), there was no interaction between treatment and plant species (= 0.890). Levels of damage by the two most common herbivore guilds, leaf-mining moths and leaf chewers, primarily larval lepidopterans and grasshoppers, were unaffected by previous CO2 treatment, for all four plant species (Figs 7, 8, P = 0.975 for leaf miners, = 0.811 for leaf chewers). Although the amount of leaf mining and leaf chewing differed between plant species (< 0.001 and = 0.001, respectively), there was no interaction between previous CO2 level and plant species, indicating that the response to previously elevated CO2 was the same across all plant species (= 0.647, leaf miners; = 0.944, leaf chewers).

Figure 6.

Foliar nitrogen concentration, ± SEM, in elevated CO2 and ambient CO2 legacy plots at Kennedy Space Center, FL, USA. Open bars, ambient CO2; closed bars, elevated CO2.

Figure 7.

Density of leaf mines per 200 leaves, ± SEM, in elevated CO2 (closed bars) and ambient CO2 (open bars) legacy plots at Kennedy Space Center, FL, USA.

Figure 8.

Density of chewed leaves per 200 leaves, ± SEM, in elevated CO2 (closed bars) and ambient CO2 (open bars) legacy plots at Kennedy Space Center, FL, USA.


Herbivore damage and biodiversity

Elevated CO2 reduced the densities of all herbivore-damaged leaves, which included damage produced by leaf miners, leaf tiers, leaf chewers and leaf gallers, on all host plant species, including the nitrogen-fixing legume, Galactia. Only pathogen damage was not consistently depressed in elevated CO2. Although longer term counts revealed that numbers of leaf miners and leaf tiers per 200 leaves in elevated CO2 were decreased in nearly all years (Stiling et al., 2009), this is the first time we have shown similar reductions for other herbivores, such as leaf gallers and for herbivory by leaf chewers. Our results are similar to those of other studies, most of which have also found reductions in insect herbivory under elevated CO2 (reviewed in Lincoln et al., 1993; Watt et al., 1995; Bezemer & Jones, 1998; Hunter, 2001; Whittaker, 2001; Stiling & Cornelissen, 2007; Lindroth, 2010; Robinson et al., 2012). Several mechanisms are responsible for this decline. First, elevated CO2 inhibits the assimilation of nitrate into organic nitrogen compounds (Bloom et al., 2010). This tissue nitrogen reduction causes reduced insect herbivore survival and reproduction. Foliar nitrogen reductions in our oaks averaged between 7% and 10% across all years, and reductions in Galactia averaged 15% (Stiling et al., 2009). Second, elevated CO2 can cause increases in allocations to carbon-based secondary metabolites, such as condensed and hydrolyzable tannins (Peñuelas & Estiarte, 1998). Earlier studies in our system showed a trend towards increased total phenolics, condensed and hydrolyzable tannins (Rossi et al., 2004; Hall et al., 2005). Third, reduced leaf quality often delays insect development (Stiling & Cornelissen, 2007; Robinson et al., 2012) and, in our system, this exposes herbivores longer to natural enemies, increasing herbivore death rates (Stiling et al., 1999), although such increases in mortality are not always evident (Lindroth, 2010).

Earlier results from our pitfall traps showed that, although there was a significant increase in herbivore catches in pitfall traps in elevated relative to ambient CO2, these increases were not evident at other trophic levels, such as insectivores, parasitoids and predators, or decomposers (Stiling et al., 2010). Because of the limited trophic cascade of CO2 from plants to other trophic levels, it is not surprising that biodiversity was not affected at the level of either insect order or beetle family. Perhaps a more detailed examination would reveal finer scale changes, but this would involve the identification of insects to family or species, which would be logistically difficult. Other studies that have examined the influence of elevated CO2 on insect biodiversity have also failed to find many significant effects (Sanders et al., 2004; Hillstrom & Lindroth, 2008). This may be because such studies have focused on species-rich communities, where reductions in some species may be offset by increases in others. Only in communities dominated by a few species might biodiversity be affected by elevated CO2 (Altermatt, 2003). However, it is possible that studies over much longer time frames would reveal changes in biodiversity. Previous studies in our system have shown increases in acorn production under elevated CO2 for Q. myrtifolia and Q. chapmanii, but not for Q. geminata (Stiling et al., 2004). Over long time periods, such effects would almost certainly affect plant diversity, and thus insect diversity, given that the different oak species support different herbivore species, albeit from the same or similar genera.

Legacy effects

There was evidence of legacy effects on fine root growth, because fine root growth in the previously elevated CO2 plots was greater than that in the ambient CO2 plots. However, levels of foliar nitrogen in previously elevated CO2 and ambient CO2 plots were statistically indistinguishable. As a result, there was no legacy of elevated CO2 on herbivory in our scrub oak forests. Shortly after the CO2 treatment ceased, herbivory increased to normal levels. The legacy of below-ground carbon accumulation and progressive nutrient limitation does not appear as important as the short-term effects of changes in foliar nitrogen. One clear implication of this is that our current generation of global change experiments may reasonably capture the dominant effects of elevated CO2, even over longer time scales than those over which we are currently capable of running experiments.

An abrupt return of atmospheric CO2 levels to ‘normal’ is no more unrealistic an analog for a future scenario than is the abrupt increase in CO2 concentrations used in typical step-change experiments. Yet, both simulations provide insight into the nature of ecosystem responses to this chronic global environmental change. Specifically, the examination of the legacy of CO2 effects in ambient conditions gauges the inertia of the ecosystem to CO2 enrichment, without the confounding influence of ongoing CO2 treatment. Thus, responses after CO2 enrichment ceases can be ascribed unequivocally to CO2-induced changes in ecosystem structure and functioning that occurred earlier, and to responses that persist beyond the cessation of CO2 exposure. The lack of substantial legacy effects provides a unique insight into critical and currently poorly understood mechanisms of ecosystem responses to elevated CO2 over cycles of disturbance and recovery. In short, we offer this analysis, not as a direct analog for future CO2 reduction scenarios, which will obviously occur on a different time scale, but rather to test hypotheses about the nature of ecosystem responses to elevated CO2. Will there be a similar lack of CO2 legacy effects in other systems? At present, we cannot be sure, because there have not been any similar studies. It is possible that legacy effects may be more likely in systems with better developed soils, higher nitrogen and a more pronounced carbon sink. However, the results of our study suggest that other fire-dependent systems dominated by perennial plants may show a similar dearth of legacy effects. How do the legacy effects of elevated CO2 compare with those of other environmental perturbations, such as acid rain or deforestation? Dobson et al. (1997) suggested a linear relationship between spatial scale of disturbance and community recovery time. In this model, the recovery time for large-scale environmental perturbations, such as acid rain and groundwater exploitation, is much longer than that of small-scale perturbations, such as tree falls and lightning strikes. This scenario might not hold for the effects of elevated CO2, where changes over large spatial scales could have few substantial legacy effects. We encourage scientists to tackle these and other questions we have raised here during the course of our studies.


This research was supported by the Office of Science (BER), US Department of Energy, through Southeast Regional Center of the National Institute for Global Environmental Change grants to P.S., by a National Science Foundation (NSF) grant to B.A.H. and by Department of Energy grants to B.D. Jamie Colson-Moon helped collect the pitfall traps. Thanks are due to Sylvia Lukasiewicz, Terri Albarricin, Kerry Bohl, Heather Jezorek, Kara Winston, Christina Harris, Georgina Johnson, Heather Faulkner, Toni Gordon, Arnaldo Villafranca, Ciro Vasquez, Samvid Owivedi, Shawn Simmons, Andy Paluch, Matt Dumouchel, Crystal Bernarducci, Shalane Ponsell, Pauline Thai, Jessica Allen, Amanda Ditson, Hamid Hoveida, Caitlyn Palmby, Carl Franconi, Craig Beatty and Dianne Harshberger for help in sorting pitfall samples and leaf litter. Ben Duval, Paul Dijkstra and Rick Doucett helped with the nitrogen analyses. We acknowledge the support and encouragement of the NASA Kennedy Space Center and Dynamac Corporation, especially Ross Hinkle. Dave Johnson, Hans Anderson, Tom Powell and Graham Hymus provided a happy working environment at the field site.