Inter-annual changes in detritus-based food chains can enhance plant growth response to elevated atmospheric CO2
Abstract
Elevated atmospheric CO2 generally enhances plant growth, but the magnitude of the effects depend, in part, on nutrient availability and plant photosynthetic pathway. Due to their pivotal role in nutrient cycling, changes in abundance of detritivores could influence the effects of elevated atmospheric CO2 on essential ecosystem processes, such as decomposition and primary production. We conducted a field survey and a microcosm experiment to test the influence of changes in detritus-based food chains on litter mass loss and plant growth response to elevated atmospheric CO2 using two wetland plants: a C3 sedge (Scirpus olneyi) and a C4 grass (Spartina patens). Our field study revealed that organism's sensitivity to climate increased with trophic level resulting in strong inter-annual variation in detritus-based food chain length. Our microcosm experiment demonstrated that increased detritivore abundance could not only enhance decomposition rates, but also enhance plant growth of S. olneyi in elevated atmospheric CO2 conditions. In contrast, we found no evidence that changes in the detritus-based food chains influenced the growth of S. patens. Considered together, these results emphasize the importance of approaches that unite traditionally subdivided food web compartments and plant physiological processes to understand inter-annual variation in plant production response to elevated atmospheric CO2.
Introduction
Anthropogenic activities currently influence ecosystems worldwide (Steffen et al., 2011). For example, increased combustion of fossil fuels and production of cement since the industrial revolution are associated with concomitant increases in atmospheric CO2 concentrations from 280 to 380 ppm, and current concentrations are projected to double by 2100 (IPCC, 2013). These increases in atmospheric CO2 are linked to warming of the earth's surfaces (Shakun et al., 2012), ocean acidification (Orr et al., 2005), and changes in global precipitation patterns (Liu et al., 2007; Solomona et al., 2009). Therefore, the principal objectives of many environmental policies have been to reduce CO2 emissions into ecosystems, and to enhance the ability of ecosystems to absorb excess CO2 (Kitzes et al., 2008).
Plants are the primary link between the atmosphere and biosphere, and rising atmospheric CO2 is expec-ted to stimulate photosynthesis and plant growth. Evidence from long-term CO2 enrichment studies suggests that plants can consistently accumulate more carbon in their biomass under elevated compared to current CO2 conditions (Rasse et al., 2005; Drake, 2014), although these effects are generally more pronounced for plant species that use C3 compared to C4 photosynthetic pathways (White et al., 2012; Erickson et al., 2013). This leads to the suggestion that highly productive ecosystems, such as wetlands dominated by C3 plants, may be a net carbon sink in future atmospheric CO2 conditions and that wetland conservation may be an effective strategy to mitigate the harmful effects of CO2 on the global climate (Mcleod et al., 2011). This suggestion has been controversial, however, and much research has focused on environmental factors such as salinity (Erickson et al., 2007), nitrogen deposition (Langley & Megonigal, 2010), and production of methane (Dacey et al., 1994; Bridgham et al., 2013), which may modify, or counteract, the value of wetlands as carbon sinks (Drake, 2014). Notably, the strength of interactions among animal consumer species may also alter the ability of plants to accumulate carbon in their biomass in the future, but we lack rigorous experimental evidence documenting the magnitude of potential consumer effects on plant growth response to elevated atmospheric CO2.
Due to their importance for litter decomposition and subsequent mobilization of nutrients that may otherwise limit plant growth (Hines & Gessner, 2012), animals in detritus-based food chains may be particularly important in elevated atmospheric CO2 conditions when plants have increased nutrient demand (De Graaff et al., 2006; Reich & Hobbie, 2013). Evaluations of the influence of elevated CO2 on the feedback between decomposition and plant growth, however, have generally focused on litter chemistry as a key rate limiting factor (Strain & Bazzaz, 1983; Norby et al., 2001), and less emphasis has been placed on the role of consumers in the detritus-based food chain. For example, initial hypotheses suggested that if the effect of elevated CO2 on litter chemistry was similar to its effects on live plant tissue [decreased nitrogen concentrations and increased concentration of lignin (Cotrufo et al., 1998)], then litter chemistry could slow decomposition, and lead to progressive limitation of plant growth in elevated atmospheric CO2 conditions (Strain & Bazzaz, 1983). However, a meta-analysis showed that efficient nutrient resorption during senescence can minimize the influence of elevated CO2 on litter chemistry (Norby et al., 2001), which reduces the likelihood that delays in decomposition due to litter chemistry will limit plant growth response to elevated CO2 in the long term. Yet, without an explicit consideration of the consequences of variation in detritivore abundance, these litter chemistry results reflect only a partial understanding of the influence of decomposition on plant growth in elevated atmospheric CO2 conditions.
At least two factors may influence the magnitude, and potentially the direction, of detritivore effects on plant growth response to elevated atmospheric CO2. First, detritivores are sensitive to desiccation, and their survival and fecundity rely on minimizing water loss (Hassall et al., 2010). Abiotic factors, such as precipitation and temperature that limit detritivore populations (Brody & Lawlor, 1984; Carefoot, 1993), also influence plant growth (Drake, 2014), but little is known about the relative magnitude of direct and detritivore-mediated effects of climate on plant growth. Second, climate may have differential effects on detritivores and the predators that consume them (MacLennan et al., 2011; Hansson et al., 2012; Sentis et al., 2013). Long-term monitoring of grassland invertebrate communities has shown that sensitivity of organisms to climatic factors, such as temperature and precipitation, can increase with trophic rank (Voigt et al., 2007). However, elevated atmospheric CO2 effects on higher trophic levels can be quite variable (Blankinship et al., 2011). Notably, predator–detritivore–litter interactions have predominately been studied in context of understanding decomposition alone (Kajak, 1995; Wardle et al., 1998; Hunter et al., 2003; Lawrence & Wise, 2004). The extent to which these effects will attenuate within the detritus-based food chain, or cascade to affect production of live plant tissue in elevated CO2 conditions, remains an open question.
We conducted a field plot and microcosm experiment to achieve two main objectives: (i) to determine whether climate influences variation in detritus-based food chains; (ii) to evaluate the effect of variation in detritus-based food chain length on the growth response of plants to elevated CO2. We hypothesize that organisms at higher trophic levels, namely spiders, will be more sensitive to variation in climate than organisms, such as their detritivorous prey that feed at lower trophic levels. Differential sensitivity of trophic levels to climate results in inter-annual variation in detritus-based food chain length. We hypothesize that variation in food chain length will have stronger influence on plant growth in elevated CO2 conditions when there is increased plant demand for nutrients that are mobilized from litter by detritivores. By comparing the influence of changes in detrititus-based food chains on decomposition and growth of two important plant functional groups (C3 vs. C4 plants), we can start to evaluate the role of consumers in determining wetland ecosystem response to elevated CO2.
Materials and methods
Study system
This study was conducted in a brackish marsh at the Smithsonian Environmental Research Center in MD, USA (38°53′N, 76°33′W). Plant communities at this site are dominated by Scirpus olneyi (a C3 sedge) and Spartina patens (a C4 grass). The species name of Scirpus olneyi has been changed to Schoenoplectus americanus (Persoon) Volkart ex Schinz and R. Keller, but throughout this manuscript, we retain the former name to maintain continuity with previous results published from the long-term open topped chamber experiment which was established at this site in 1987 (Drake et al., 1989). Results to date show that S. olneyi has a sustained increase in primary production under elevated CO2 conditions (Rasse et al., 2005). In contrast, the biomass of S. patens is less affected by elevated atmospheric CO2 conditions (Arp et al., 1993; Drake, 2014).
The numerically dominant predator, Pardosa littoralis (Araneae: Lycosidae), forages broadly across S. olneyi and S. patens communities. These mobile spiders also aggregate in areas with abundant leaf litter where they feed on a broad variety of invertebrate prey including detritivorous isopods, such as Littorophiloscia vittata (Isopoda: Philosciidae) (Hines & Gessner, 2012). This isopod lacks pleopodal lungs, making it poorly adapted to absorbing moisture from the air. Consequently, individuals often aggregate in high densities, which creates more favorable microclimate for the group and intensifies their influence on mineralization of leaf litter (Hassall et al., 2010; Hines & Gessner, 2012).
Field experiment
To examine the influence of climate on species in detritus-based food chains, we sampled 0.47 m2 field plots containing either S. olneyi or S. patens communities. The open field plots were originally established in 1987 as part of a long-term study examining the effects of elevated CO2 on plant productivity (Drake et al., 1989). Description of the plots and long-term monitoring of plant biomass are described in detail by Erickson et al. (2007) and Drake (2014). Here, we focus on the invertebrate sampling, which is reported for the first time. Invertebrates were sampled five times over 6 years (28 June 2004, 20 July 2005, 19 July 2006, 23 July 2007, 27 July 2009), an appropriate temporal resolution because the focal species show relatively low intra-seasonal variation (Hines & Gessner, 2012). On each date, we estimated densities of focal arthropods (L. vittatta and P. littoralis) using a standard suction sampling method consisting of three, 3-second vacuum suction samples taken with a d-vac® vacuum sampler (Ventura, CA, USA) fitted with a 0.20 m2 nozzle (Hines et al., 2006). Arthropods were killed in the field with ethyl acetate vapor before they were counted in the laboratory.
The effect of atmospheric CO2 on abundance of P. littoralis spiders and L. vittata isopods was assessed using repeated measures anova with plant community (S. olneyi and S. patens) considered as a categorical effect and time considered as a continuous effect. Animal densities were log-transformed to meet assumptions of normality and homogeneity of variances for analysis, and untransformed densities are shown in the figures. To estimate sensitivity of predators, detritivores, and plants to variation in climate, we collected five climatic variables reported at a local weather station (Table 1). The chosen variables reflect overall climate conditions, and climatic stress due seasonal temperature extremes, and water stress. To account for potential correlation among climate variables, we simplified the data into two composite principal component axes, which explained 89% of the variation in the climate data. The fraction of temporal variance in animal density and plant biomass that could be accounted for by two composite climatic variables was estimated using partial redundancy analysis (pRDA), which is a multivariate analog of regression (R package Vegan) (Hammock & Johnson, 2014).
| dfa | Spider density | Isopod density | Predator: prey | ||||
|---|---|---|---|---|---|---|---|
| F | p | F | p | F | p | ||
| Plant | 1,8 | 1.17 | 0.31 | 3.17 | 0.11 | 1.55 | 0.24 |
| Year | 1,38 | 9.04 | 0.005 | 0.89 | 0.35 | 10.76 | 0.002 |
| Plant × Year | 1,38 | 1.80 | 0.19 | 3.92 | 0.05 | 1.19 | 0.28 |
- a Degrees of freedom: numerator, denominator.
Microcosm experiment
To compare the influence of detritus-based food-chain length on litter mineralization and subsequent wetland plant growth of two plant species (S. olneyi and S. patens) under elevated as opposed to ambient atmospheric CO2 concentrations, we conducted a 2 × 4 factorial microcosm experiment with a randomized complete block experimental design. Treatment factors included two levels of atmospheric CO2 (ambient atmospheric CO2: 365 ppm, or elevated atmospheric CO2: 705 ppm) and four food-chain length (FCL) treatments: (i) control-litter withheld, (ii) litter addition (5 g per microcosm corresponding to 525 g m−2), (iii) litter and 10 L. vittata isopod addition (corresponding to 1275 isopods m−2), and (iv) litter, isopod and 2 P. littoralis spider addition (corresponding to 250 spiders m−2). Due to clumped spatial distribution in the field, ten isopods at the base of a plant is reasonable, if not a slightly lower density than natural aggregations. Nonetheless, these initial species densities were chosen to fall within the range of naturally heterogeneous animal densities in the field, to allow for mortality of animals during experimental setup, and to stay within the carry capacity of the microcosms. Treatments involving predators without prey were not included because starvation of animals for the duration of the experiment would be unnecessarily cruel. All treatments were replicated six times for a total of 96 experimental microcosms (2 CO2 treatments × 4 food-chain length treatments × 2 plant species × 6 replicates).
Atmospheric CO2 treatments were established in outdoor chambers (1.5 m long × 1 m wide × 1.5 m tall, wooden enclosures surrounded by polyester film) as described by Wolf et al. (2007). Detritus-based food chain treatments were established in microcosms placed within CO2 chambers. Each microcosm consisted of a sand-filled pot (11 cm diameter × 16 cm deep), inoculated with a slurry of natural microbial community and exposed to constant moisture by seating them within pots (16 cm diameter × 16 cm deep) filled to 12 cm with deionized water. Each microcosm contained either S. olneyi grown from field-collected, preweighed rhizome nodes, or S. patens, which was propagated from field-collected seeds and grown to the height of 5 cm before the start of the experiment on May 22, 2007. Litter addition treatments were established using leaf litter collected from S. olneyi or S. patens plants that were grown and allowed to completely senesce under either elevated or ambient conditions in the above chambers during the previous year. Clear plastic tube cages of 10 cm diameter × 1 m tall were sunk in to the soil to enclose the plants and litter community in each microcosm.
To assess invertebrate survival, decomposition, and plant growth through time, half of the experimental units were harvested on each of two sample dates (day 50 and 84, corresponding to July 11, 2007 and August 14, 2007, respectively). On each sampling date, surviving invertebrates were counted following careful visual inspection of microcosms and sieving of soil. Litter remaining, as well as live root and shoot tissue for each plant species, was harvested, washed, and dried at 50°C before each tissue was weighed. Litter mass loss rate coefficients (k) were calculated using an exponential decay model (Mt = M0 e−kt), where Mt = mass at time t, M0 = initial litter mass, and t = elapsed time in days. All data met assumptions of homogeneity of variance and normality of residuals. Therefore, the main and interactive effect of atmospheric CO2 and detritus-based food chain treatments on response variables (isopod abundance, litter mass loss rate, root biomass, shoot biomass) was assessed using mixed-model anova. Categorical factors included atmospheric CO2 (elevated or ambient) and detritus-based food chain treatments (control, litter addition, isopod addition, isopod and spider addition), and time was considered a continuous factor for the analysis of animal abundance and live plant biomass. Microcosms were considered independent experimental units because they were not hydrologically connected, and variance associated with chambers was effectively zero.
Results
Field experiment
There was strong inter-annual variation in the density of predatory spiders as well as the ratio of predators to their isopod prey (Fig. 1; significant year effects: Table 1), which was partly explained by climate (Fig. 2; Table 2). The pRDA showed that effects of climate were consistently stronger on higher trophic levels in both plant communities (Fig. 2). That is, the two principle component climate axes explained 70–73% of the variation in spider density, 20–21% of the variation in isopod density, and 0–30% of the variation in S. olneyi and S. patens plant biomass, respectively (Fig. 2).
| Figure 2 Label | Description | PCA1 | PCA2 |
|---|---|---|---|
| X1 | Mean air temperature, whole year | −0.452 | |
| X2 | No. days with maximum temperature >30°C | −0.534 | −0.163 |
| X3 | No. days with minimum temperature <0°C (preceding winter) | 0.183 | 0.879 |
| X4 | No. days incident light <40 mol quanta m−2 day−1 | 0.510 | −0.191 |
| X5 | Total precipitation May–July (mm) | −0.465 | −0.398 |
- The first two axes explain 89% of total variance. Loadings >0.45 are shown in bold for clarity.
Microcosm experiment
Spiders persisted in the microcosms for the duration of the experiment (85 ± 5% survival), with some low initial mortality resulting in significant temporal effects on predator survival (Table 3); nonetheless, predator mortality was not influenced by atmospheric CO2 or plant community. Similarly, isopod mortality was low, and 75–80% of the isopods survived in the absence of predation (Fig. 3). Spiders consumed isopods, but may have been satiated over the course of the experiment as predation on isopods was greater during the first 50 days of the experiment (Fig. S1), and prey populations were not reduced further by day 84 (Fig. 3; Table 4). Overall, spiders suppressed isopod populations by 55% compared to isopod populations in spider-free treatments (Fig. 3). Predation was equally effective in both plant communities and was not influenced by atmospheric CO2 conditions (Table 4, Fig. 3).
| Scirpus olneyi | Spartina patens | ||||
|---|---|---|---|---|---|
| dfa | F | p | F | p | |
| CO2 | 1,20 | 0.0 | 1.00 | 0.00 | 1.00 |
| time | 1,20 | 15.9 | <0.001 | 8.5 | 0.009 |
| CO2 × time | 1,20 | 0.00 | 1.00 | 0.24 | 0.63 |
- a Degrees of freedom: numerator, denominator.
| Scirpus olneyi | Spartina patens | ||||
|---|---|---|---|---|---|
| dfa | F | p | F | p | |
| CO2 | 1,40 | 0.42 | 0.52 | 0.76 | 0.39 |
| Predation (FCL) | 1,40 | 27.2 | <0.001 | 29.2 | <0.001 |
| time | 1,40 | 121.8 | <0.001 | 67.1 | <0.001 |
| CO2 × FCL | 1,40 | 0.047 | 0.083 | 0.49 | 0.49 |
| CO2 × time | 1,40 | 0.86 | 0.36 | 0.16 | 0.70 |
| FCL × time | 1,40 | 24.1 | <0.001 | 25.0 | <0.001 |
| CO2 × FCL × time | 1,40 | 0.10 | 0.75 | 0.68 | 0.42 |
- a Degrees of freedom: numerator, denominator.
Decomposition of both plant species was sensitive to changes in detritivore food chain length, but was not influenced by atmospheric CO2 (Table 5). That is, decomposition was accelerated by isopods, and there was less litter mass remaining when isopods were present compared to when they were absent (Fig. 4). However, spider predation, which limited isopod abundance (Fig. 3), did not have extended effects on decomposition (Fig. 4).
| Litter mass | Shoot biomass | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| dfa | Scirpus olneyi | Spartina patens | dfa | Scirpus olneyi | Spartina patens | |||||
| F | p | F | p | F | p | F | p | |||
| CO2 | 1,60 | 2.8 | 0.10 | 1.0 | 0.3 | 1,80 | 1.3 | 0.3 | 0.7 | 0.4 |
| FCL | 2,60 | 13.7 | <0.001 | 13.8 | <0.001 | 3,80 | 1.7 | 0.18 | 0.9 | 0.5 |
| Time | 1,60 | 620.1 | <0.001 | 376.4 | <0.001 | 1,80 | 249.4 | <0.001 | 79.9 | <0.001 |
| CO2 × FCL | 2,60 | 0.7 | 0.5 | 0.2 | 0.8 | 3,80 | 5.1 | 0.03 | 0.2 | 0.9 |
| CO2 × Time | 1,60 | 3.4 | 0.07 | 1.6 | 0.2 | 1,80 | 3.3 | 0.07 | 0.2 | 0.6 |
| FCL × Time | 2,60 | 12.8 | <0.001 | 13.7 | <0.001 | 3,80 | 2.3 | 0.09 | 1.8 | 0.2 |
| CO2 × FCL × Time | 2,60 | 0.5 | 0.6 | 0.2 | 0.8 | 3,80 | 1.5 | 0.2 | 0.8 | 0.5 |
- a Degrees of freedom: numerator, denominator.
There was an interactive effect of elevated CO2 and detritus-based food chain on shoot biomass of S. olneyi (Table 5, Fig. 5). In elevated CO2 conditions, the addition of isopods was associated with increased S. olneyi biomass, an effect that was not found in ambient atmospheric CO2 conditions. This positive association between L. vittata isopods and plant growth was found in the presence and absence of spiders (Table 5, Fig. 5). In contrast, variation in S. patens biomass was not explained by atmospheric CO2 conditions or detritivore food chain treatments (Fig. 5, Table 5).
Discussion
Our 6-year field study shows that an organism's sensitivity to climate increases with its trophic level. In particular, there were strong fluctuations in the relative abundances of P. littoralis spiders and L. vittata isopods, two numerically dominant species in our detritus-based food chain. This result is consistent with longer-term monitoring studies, which examined larger numbers of species and climate factors (Voigt et al., 2007; Hammock & Johnson, 2014), lending support that our conclusions are not biased by the particular species or climate factors that we included in our analyses. Consequently, it is likely that similar inter-annual variation in detritus-based food chains has the potential to influence decomposition in a wide variety of ecosystems.
Our microcosm study shows that variation in detritus-based food chain can have important influences on two important ecosystem processes in wetland ecosystems. Not only did increased L. vittata abundance accelerate decomposition, it was also associated with enhanced production of S. olneyi biomass in response to elevated CO2. Specifically, in the presence of litter and isopods, elevated CO2 enhanced S. olneyi shoot growth by 24%, and the effect of elevated atmospheric CO2 on plant growth was weaker in control microcosms where isopods and litter were absent (Fig. 5). The magnitude of shoot growth response to elevated CO2 and isopods is consistent with the influence of CO2 on S. olneyi growth in the field, which can range from 0 to 37% depending on variation in environmental conditions (Drake, 2014). The interactive effect of detritivore food chain and elevated atmospheric CO2 on plant growth that we found was plant species specific (Table 5), whereby S. olneyi shoot biomass was enhanced, but no equivalent effect was found for S. patens. These results are consistent with our predictions that mineralization of leaf litter by detritivores in elevated CO2 conditions would be more important for plants with C3 photosynthetic pathway. A few considerations about linkages between detritus-based food chains and plant production, however, warrant further discussion.
First, spiders consumed isopods in microcosms, but they did not limit the positive effect of isopods on litter decomposition or S. olneyi growth, which demonstrates that trophic interactions can be surprisingly complex, even in tractably simplified experimental food cha-ins. One explanation is that there may be a temporal component to isopod effects on plant growth, whereby isopods mineralized leaf litter, which stimulated S. olneyi growth before the isopods were consumed by spiders. That detritivore-mediated effects on plant growth were greater on the first (Fig. S2) compared to the second harvest date (Fig. 5) lends some support to the temporal hypothesis.
Decomposition rate, however, was similar in all microcosms containing isopods (Fig. 4) despite reductions in isopod densities when isopods and spiders were present in combination (Figs 3 and S1). Therefore, it is likely that that per capita feeding rate of L. vittata isopods on leaf litter was higher when P. littoralis spiders were present, which suggests that higher order (consumer–consumer) interactions may be playing an important role in determining the effects of elevated atmospheric CO2 on ecosystem processes. Such effects could be due to enhanced isopod metabolism resulting from fear of predation as has been demonstrated in grasshopper–spider interactions (Hawlena & Schmitz, 2010), or due to reductions in interference competition among isopods at lower densities, which would allow for increased foraging time for consumers (i.e. McPeek, 1998). Alternatively, it is possible that spider excretions and decomposition of exsanguinated isopod bodies could have priming effects on soil microbial communities with extended effects on decomposition that rivaled in magnitude the direct consumptive effects of litter by detritivores (MacLennan et al., 2011). This is not an unrealistic possibility as similarly dramatic priming effects have been demonstrated following decomposition of grasshoppers (Hawlena et al., 2012).
Although we are not able to isolate the specific interaction pathway by which spider–isopod interactions influence decomposition, our results effectively demonstrate that consumer interactions in the detritus-based food chain influenced S. olneyi decomposition, which was also associated with enhanced S. olneyi growth response to elevated CO2. An important caveat is that, contrary to expectation, addition of leaf litter and subsequent feeding by isopods did not result in enhanced shoot biomass in ambient CO2 conditions for either plant species. This suggests that factors other than nutrients release from litter limited plant growth in these treatments and highlights the importance of context dependency of this interaction pathway. By focusing our microcosm experiment on four different food chain treatments and two abiotic environmental conditions (elevated and ambient atmospheric CO2), we limited our ability to assess how consistent detritus-based food chain effects will be across a range of climate conditions, such as changes in temperature, salinity, and precipitation. We see much potential for future experiments that follow this line of inquiry.
In conclusion, we find that primary producers and higher trophic levels show differential sensitivity to the effects of inter-annual climate variation and elevated atmospheric CO2. Although detritivores can be patchily distributed in the field, where present, the magnitude of their influence on litter decomposition and also on shoot biomass can rival in magnitude the interactive effects of elevated CO2 and other factors of environmental variation, such as precipitation and increased nitrogen deposition (Erickson et al., 2007; Reich & Hobbie, 2013; Reich et al., 2014). While conservation of habitats that support detritivore consumers may influence the ability of wetlands to absorb some atmospheric CO2, we do not overlook the potential for other factors, such as methane emissions (Bridgham et al., 2013) or competitive exclusion of productive plants (Langley & Megonigal, 2010) to counteract effective carbon sequestration in wetlands. Instead, our results emphasize the importance of approaches that unite traditionally subdivided food web compartments and biogeochemical processes to maximize predictive understanding of ecosystem response to elevated CO2.
Acknowledgements
We thank Yuri Mori, Hayes Biche, and Gary Peresta for help in the field, John Erickson and members of the crab laboratory for providing EFS, and Meng Lu for providing climate data. This work was funded by EPA-STAR fellowship FP-91648701-1 to JH. Further support came from the German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, funded by the German Research Foundation (FZT 118).
