Don’t drink it, bury it: comparing decomposition rates with the tea bag index is possible without prior leaching

The standardized ‘Tea Bag Index’ enables comparisons of litter decomposition rates, a key component of carbon cycling, across ecosystems. However, tea ‘litter’ may leach more than other plant litter, skewing comparisons of decomposition rates between sites with differing moisture conditions. Therefore, some researchers leach tea bags before field incubation. This decreases comparability between studies, and it is unclear if this modification is necessary. We submerged green and rooibos tea bags in water, and measured their leaching losses over time (2 min – 72 h). We also compared leaching of tea to leaf and root litter from other plant species, and finally, compared mass loss of pre-leached and standard tea bags in a fully factorial incubation experiment differing in soil moisture (wet and dry) and soil types (sand and peat). Both green and rooibos tea leached strongly, levelling-off at about 40% and 20% mass loss, respectively. Mass loss from leaching was highest in green tea followed by leaves of other plants, then rooibos tea, and finally roots of other plants. When incubated for 4 weeks, both teas showed lower mass loss when they had been pre-leached compared to standard tea bags. However, these differences between standard and pre-leached tea bags were similar in moist vs. dry soils, both in peat and in sand. Thus, despite large leaching losses, we conclude that leaching tea bags before field or lab incubation is not necessary to compare decomposition rates between systems, ranging from as much as 5% to 25% soil moisture.


Introduction
Soils store more than three times the amount of carbon than what is currently present in the atmosphere (Fischlin et al. 2007). This carbon storage is heavily controlled by decomposition of organic material, and accurately assessing decomposition rates in terrestrial ecosystems is therefore important to predict current and future soil carbon storage. A very common and useful technique for this assessment are litterbags; small mesh bags filled with plant material which are incubated in the field. However, comparisons of study results across local or global scales have historically been difficult due to differences in the plant and bag material used, as well as the general methodology (depth of burial, drying of litter before incubation, mesh size of litter bags, etc.). To facilitate large-scale comparisons between studies Keuskamp et al. (2013) developed the 'Tea Bag Index' which uses commercially available tea bags and a standard protocol. Indeed, the Tea Bag Index has been widely implemented since its introduction both in traditional scientific and crowd-sourced science studies (e.g., Djukic et al. 2018). In addition, tea bags have also increasingly been used as a readily-available standard material for comparisons of decomposition between sites or treatments within the same study and over time (e.g., Elumeeva et al. 2018;Marley et al. 2019;Schwieger et al. 2020).
Despite the many advantages of using tea bags as standard material, there may also be some inherent problems with this method. It seems likely that plants that are used for tea have a particularly high amount of water-soluble compounds to ensure strong flavour. That is, compared to other plants tea might leach (i.e., lose weight due to loss of water-soluble compounds) disproportionally more, especially in soils with high water contents. Differences in decomposition between sites that are driven by precipitation or soil moisture content (Djukic et al. 2018;Petraglia et al. 2019) may thus only be present in material susceptible to high leaching, such as tea, and not in other plant material. If true, this could lead to erroneous conclusions about drivers of decomposition rates between sites. As a solution, it has been suggested to leach tea bags before they are brought out into the field (e.g., Pouyat et al. 2017;Seelen et al. 2019). Yet, to our knowledge, it has not been tested if green and rooibos tea does indeed leach more than other plant material and if those differences could influence results of decomposition rates under different moisture regimes. Undoubtedly, comparability between studies would decrease substantially if the tea bag protocol is adapted to include leaching before incubation in some but not all studies, and at present, it is unclear if this modification is necessary.
Here, we studied if leaching of tea bags before field incubation is needed to meaningfully compare differences in decomposition rates across habitats of differing moisture conditions. Thereto, we (i) constructed leaching curves for both green and rooibos tea to assess at which point leaching levels-off, then (ii) compared leaching of green and rooibos tea to leaf and root material from three other plant species (a tree, a herb, and a grass), and finally, (iii) compared mass loss of pre-leached and non-leached tea bags in a fully factorial incubation experiment differing in soil moisture (wet and dry) and soil types (sand and peat).

Tea-and litterbags
For all tea bags, we used commercially available Lipton tea bags (green Camellia sinensis, EAN no.: 8722700 055525, and rooibos tea Aspalanthus linearis, EAN no.: 8722700 188,438) made of nonwoven polypropylene in a tetrahedron shape. Additional litterbags (nylon, 5 cm × 6 cm, mesh size 0.14 mm) were filled with leaf and root material from the tree Alnus glutinosa (L.) Gaertn. (black alder), the herb Lythrum salicaria L. (purple loosestrife) and the grass Deschampsia cespitosa (L.) Beauv. (tufted hairgrass). Alnus glutinosa seedlings were 3 years old at sampling, and were obtained from a local nursery in summer 2018, and individuals of L. salicaria and D. cespitosa were collected in the field close to Greifswald in autumn of 2018. All species were kept in pots in the common garden at the Institute of Botany and Landscape Ecology at the University of Greifswald, Germany, until harvest in June 2019. Leaves and fine roots (<2 mm diameter) of each species were harvested, washed clean, oven-dried at 70°C for 48 h, and cut to a maximum size of 1 cm 2 . For each species, we filled three replicate litterbags with roots and leaves, recorded the exact weight of the material which was always 0.5 g, and heat-sealed them. We then used these tea and litter bags in the different experiments, see Fig. 1 for a schematic overview. The materials of the commercial green and rooibos tea bags and bags used for additional plant litter bags unfortunately differed due to practical reasons (availability of heat-sealable material for additional bags). However, even though the non-woven Lipton bags appear to have a smaller mesh size, water soluble compounds can leave these bags easily as well (a prerequisite for tea brewing) and we therefore assume no difference in leaching from the bags of different types. This assumption may not apply to assessing decomposition rates in a field setting in which break-down by soil mesofauna is included, but we believe it to be appropriate when assessing differences in leaching through submersion in water. Further, no chemical analyses were done on the leached substrates, so interference from potential differences in of bag material was not an issue in our study.

Tea leaching curves
We soaked green and rooibos tea bags for varying time intervals to establish a leaching curve showing the relationship between biomass loss and soaking time. We assumed a non-linear relationship between mass loss and time with the strongest differences in the beginning. Thus, we used a gradient design with increasing time intervals and n = 1 per time interval, because this set-up outperforms replicated design in detecting and quantifying non-linear responses (Kreyling et al. 2018). In total, we soaked 17 tea bags of both rooibos and green tea, each one in an individual beaker filled with 500 ml of deionized water at a temperature of 25°C. The time intervals were 2 min, 4 min, 6 min, 10 min, 15 min, 25 min, 35 min, 1 h, 2 h, 3 h, 4 h, 6 h, 12 h, 24 h, 48 h, 69.7 h, and 72.4 h. Afterwards, tea bags were air-dried, oven-dried at 70°C for 48 h and weighed. A leaching curve of mass loss as a function of leaching time was established following a Michaelis-Menten equation: ỹ (Vmax*x)/(K m + x); where y is the mass loss (%), Vmax is an asymptote which represents the maximal mass loss of y, x is the leaching time and Km is the Michaelis constant which describes the input (x) that gives 50% of Vmax output.

Comparison of leaching between green and rooibos tea, leaves, and roots
We compared leaching of green and rooibos tea with leaves and roots from different taxa (see above). All bags were soaked in 500 ml deionized water at a temperature of 25°C for 12 h, air-dried, oven-dried at 70°C for 48 h, and weighed. For green and rooibos tea, we used 40 replicate bags for each tea type, for other plant material three replicate bags for each species and plant part (leaves or roots).

Fully factorial incubation experiment
For our incubation experiment, 40 standard and 40 preleached tea bags of each tea type were weighed and buried for 4 weeks (June 24th to July 17th), in separate pots (10 cm 3 ) with one of two different soil types (peat and sand) and two soil moisture conditions (wet and dry), a total of 160 samples. We prepared pre-leached tea bags similarly to the comparison of tea with leaves and roots (see above). Peat soil was standardised bog peat (Torfwerk Moorkultur Ramsloh, Werner Koch GmbH & Co. KG, Saterland, Germany). The peat was not fertilized, but limed to pH 5.6-6.4 at the time of production in May 2019, see Table S2 for more details on the peat substrate. The sand was predominantly medium and fine sand (59% and 21%, respectively, with an additional 11% coarse sand, 4% gravel, and 6% silt and clay). For each combination of treatments, we had ten replicates. Green and rooibos tea that had undergone the same treatment were buried pairwise with a distance of c. 5 cm between them, at a depth of 2 cm in pots. Wet pots were watered to maintain a soil moisture of around 0.25 m 3 /m 3 , dry pots were not watered during the experiment. All pots were randomly placed in bigger boxes in a climate chamber at 25°C and the position of the boxes was changed at least twice a week. During the first week, soil moisture was measured every day with Decagon Em 50 soil moisture sensors, and in the following weeks, it was measured two to three times a week. Pots in the 'wet' treatment were watered to Fig. 1 Overview of the experiments. In Experiment 1 we constructed leaching curves for green and roiboos tea over time, in Experiment 2 we compared leaching between green and roiboos tea, leaves, and roots, and Experiment 3 was a fully factorial incubation experiment of green and rooibos tea in sand and peat under wet and dry conditions maintain soil moisture levels. Soil moisture was higher in the wet soil treatments than in the dry soil treatments (wet soils 0.217 ± 0.003 m 3 /m 3 , dry soil 0.052 ± 0.002 m 3 /m 3 , mean ± SE; Fig. S1, Table S1), though wet peat soil (0.247 ± 0.003 m 3 /m 3 ) was wetter than wet sand (0.187 ± 0.003 m 3 /m 3 ). Soil moisture did not differ between pre-leached and standard tea bags (standard 0.135 ± 0.004 m 3 /m 3 , pre-leached 0.134 ± 0.004 m 3 / m 3 ). After 4 weeks, the tea bags were collected, airdried, cleaned, oven-dried at 70°C for 48 h, and weighed. We recorded mass loss, and calculated the decomposition rate 'k' and stabilisation rate 'S' according to the Tea Bag Index protocol (Keuskamp et al. 2013).

Statistical analysis
We pooled observations from the three 'other' plant species into the groups 'leaves' and 'roots', and quantified differences in leaching between leaves, roots, and green and rooibos tea with a linear model and a type III ANOVA (we used type III because of large differences in observations between groups). Differences among groups were subsequently tested using least square means. We quantified the difference between standard and pre-leached tea in each soil type and soil moisture condition with a linear model and type II ANOVA. We performed separate tests for the two tea types, because we were not interested in the large differences between tea types (as tested above). Power analyses were run to test that our experimental design could pick up on small (<5 percentage points) differences in our treatments (see Supporting information for details). All statistical analyses were performed in R (version 3.6.2), and packages used were 'nlme' (Pinheiro et al. 2018) and 'LmerTest' (Kuznetsova et al. 2017) for statistical analysis, 'lsmeans' (Lenth 2016) for posthoc comparison, 'minpack.lm' for fitting the Michaelis-Menten curve (Elzhov et al. 2016), 'Superpower' (Lakens and Caldwell 2021) for power analyses, and 'ggplot2' (Wickham et al. 2016) for visualizations.

Tea leaching curves
For both tea types, mass loss rapidly increased with leaching time in the beginning and then levelled off.
The fitted Michaelis-Menten equation showed that half of the mass loss from leaching already occurred at 2.4 min and 1.2 min for green (K m = 0.04) and rooibos tea (K m = 0.02), respectively (Fig. 2). Leaching losses then levelled off on a higher level for green (V max = 39.65,) than rooibos tea (V max = 17.36).

Fully factorial incubation experiment in soil
Pre-leached tea bags showed a much lower mass loss during the incubation experiment than the standard tea bags (43.7% and 62.3% for green tea, and 11.9% and 24.4% for rooibos tea, Table 1). Mass loss was also influenced by soil type, with slightly higher losses in peat than sand, for both green (55.5% in peat, 50.1% in sand) and rooibos tea (21.4% in peat, 14.9% in sand). Soil moisture condition also influenced mass loss with higher mass losses in wet than dry soils (rooibos tea: 20.0% in wet soils, 16.4% in dry soils), albeit with smaller differences than soil type and only marginally significant so for green tea (54.8% in wet soils, 51.2% in dry soils, P = 0.064). Only green tea had higher differences in mass loss between pre-leached and standard tea in sand than in peat (significant interaction between preleaching treatment and soil type; (Fig. 4, Table 1). The differences in mass loss between soil moisture conditions did depend on soil type as both tea types showed larger differences between wet and dry soil conditions in peat soil than in sand.
However, there was no significant interaction between treatment of the tea bags and soil moisture conditions, despite sufficient statistical power for detecting ecologically meaningful interaction effects (see Supplementary Information). That is, differences in mass loss between wet and dry soils did not depend on whether pre-leached or standard tea bags were incubated. There was also no significant three-way interaction between treatment, soil type and soil condition.
The patterns observed for mass loss were the same for the decomposition rate k and the stabilisation factor S (Fig. S2, Table S3). Here, we also found an effect of pre-leaching treatment and soil type, as well as an interaction of soil type and soil moisture condition in both decomposition rate k and stabilisation factor S. Decomposition rates were also slightly higher in wet soils than dry soils, and the stabilisation factor showed larger differences between pre-leached and standard tea bags in sand than in peat (significant interaction between treatment and soil type). But again, neither differences in decomposition rate k nor in stabilisation factor S between wet and dry soils depended on whether pre-leached or standard tea bags were used (no significant interaction between pre-le aching t rea tme nt a nd soil m oisture conditions).  Fig. 3 Mass loss of green and rooibos tea, as well as leaves and roots (from * Alnus glutinosa,°L ythrum salicaria, and + Deschampsia cespitosa) after 12 h of leaching. Shown are boxplots with the median, and in which the lower and upper hinges correspond to the 25th and 75th percentiles, and the whiskers extend from the hinges to the values closer than 1.5 × the interquartile range, n = 40 for green and rooibos tea, n = 3 for leaves and roots. Each symbol represents one sample

Discussion
We used a suite of experiments to explore if preleaching, 'brewing', of tea bags is necessary before incubation to accurately assess differences in decomposition between sites or ecosystems that differ in soil moisture. As expected, both tea types did leach substantially upon submergence in water. Nevertheless, while green tea did leach almost twice as much as leaves and almost four times as much as roots, leaching losses from rooibos tea were broadly similar to other leaf material. We also showed that despite lower mass losses when tea bags were pre-leached, the relative differences were similar between wetter and drier soil conditions.
Tea leaches quickly and substantially, but other plants do too Both tea types showed a fast and substantial mass loss in response to water inundation. This is not surprising, as Keuskamp et al. (2013)  When we compared leaching losses of tea to leaves and roots of three temperate plant species (A. glutinosa, L. salicaria, and D. cespitosa), we saw that while green tea did leach about double the amount, rooibos tea leached on average slightly less than the leaves of the selected plants. This underlines that tea is indeed suitable as a proxy for other plant material (Didion et al. 2016;Duddigan et al. 2020). Though the focus of decomposition studies generally is on leaf litter, we want to point out that root litter did leach substantially less in our experiment than plant leaf litter (including both tea types). In many ecosystems, such as shrublands, grasslands, or tundra, belowground biomass exceeds aboveground biomass (Mokany et al. 2006), highlighting the important role that root litter may have in ecosystem carbon fluxes and storage. This does caution against the representation of tea as root litter, which should be kept in mind when following the instructions of the Tea Bag Index and burying the tea bags at 8 cm depth (Keuskamp et al. 2013).
Mass losses from pre-leached teas are lower than standard tea bags but unaffected by soil moisture As expected, mass losses were lower from pre-leached than standard tea bags in our incubation experiment, most likely because they had already lost most watersoluble components before the start of the experiment. Rooibos tea also showed lower mass losses in drier soils. However, to our surprise, the differences between pre-leached and standard tea bags were similar in moist and dry soils, despite sufficient statistical power to detect meaningful interactions (power analysis in the Supplementary Information). That is, the difference in % mass loss between standard and pre-leached tea bags was, for green tea, 18.9 in dry soils and 18.2 in wet soils; and for rooibos tea 12.3 in dry soils and 12.9 in wet soils.
For green tea, the differences in mass loss between pre-leached and standard tea bags were higher in sand than in peat. This could be due to lower microbial activity in the sandy substrate (indicated by generally lower mass losses in sand) leading to a slightly more obvious leaching signal. This is also in line with the lack of this interaction in rooibos tea, which generally showed lower leaching losses. The superpower of the Tea Bag Index is to use readilyavailable standard material that allows comparisons from experiments around the world (e.g., Djukic et al. 2018), even implemented by crowd-sourced science. However, this advantage of comparability quickly degrades if different pre-treatments are introduced. Our results show that, at least in the broadly different soil moisture regimes tested here, the aims of the Tea Bag Index to serve as a standard material with which environmental influences on decomposition can be compared across sites are not compromised.