Antibiotic‐induced effects on scaling relationships and on plant element contents in herbs and grasses

Abstract Plant performance is correlated with element concentrations in plant tissue, which may be impacted by adverse chemical soil conditions. Antibiotics of veterinary origin can adversely affect plant performance. They are released to agricultural fields via grazing animals or manure, taken up by plants and may be stored, transformed or sequestered by plant metabolic processes. We studied the potential effects of three antibiotics (penicillin, sulfadiazine, and tetracycline) on plant element contents (macro‐ and microelements). Plant species included two herb species (Brassica napus and Capsella bursa‐pastoris) and two grass species (Triticum aestivum and Apera spica‐venti), representing two crop species and two noncrop species commonly found in field margins, respectively. Antibiotic concentrations were chosen as to reflect in vivo situations, that is, relatively low concentrations similar to those detected in soils. In a greenhouse experiment, plants were raised in soil spiked with antibiotics. After harvest, macro‐ and microelements in plant leaves, stems, and roots were determined (mg/g). Results indicate that antibiotics can affect element contents in plants. Penicillin exerted the greatest effect both on element contents and on scaling relationships of elements between plant organs. Roots responded strongest to antibiotics compared to stems and leaves. We conclude that antibiotics in the soil, even in low concentrations, lead to low‐element homeostasis, altering the scaling relationships between roots and other plant organs, which may affect metabolic processes and ultimately the performance of a plant.

For example, metabolically active leaves show higher N:P ratios than supporting structures like stems and are generally richer in P, K (potassium), and S (sulphur) than stems, whereas stems show higher C:N ratios than other plant organs (Abrahamson & Caswell, 1982;De Deyn, Cornelissen, & Bardgett, 2008;Güsewell, 2004).
Across and within plant organs, different elements are highly correlated. Kerkhoff, Fagan, Elser, and Enquist (2006) showed that scaling relationships (i.e., trends of allometry or trade-offs between physical or chemical properties, like mass, size and concentrations) of elements change between organs. The authors focused on N and P content and applied the term "structural" to group stems and roots, and "metabolic" to group leaves and diaspores. Scaling of plant elements among organs of one group (i.e., among stems and roots, or among leaves and diaspores, respectively) yielded isometric relationships (slopes ~1, i.e., constant increase of elements in compared plant organs within one group), for example root N versus stem N or leaf P versus diaspore P. Scaling relationships between plant organs of the separate groups were anisometric (for example, stem N vs. diaspore N or root P vs. leaf P). Anisometric scaling describes a stronger (or less) strong increase in an element of one organ compared to that of another organ. The authors concluded that there was a common set of rules in the partitioning of elements among plant organs, independent of plant functional groups (here: woody species and herbs).
It is notable that internal plant element concentrations are dependent on the external availability of elements (Frost, Evans-White, Finkel, Jensen, & Matzek, 2005). For example, as response to high external nutrient supplies, primary producers may maximize protein synthesis and growth as competitive strategy, which results in increased biomass element:C ratios (Ågren, 2008;Matzek & Vitousek, 2009). On the other hand, growth and biomass element:C ratios are decreased and nutrients are more effectively used by plants in response to low nutrient conditions (Vitousek, 1982).
Antibiotics in the environment have been recognized as a serious threat to nontarget organisms as well as the entire ecosystem, and have been grouped, together with other pharmaceuticals and personal-care products, in a new group of chemicals termed "contaminants of emerging concern" (Bartrons & Peñuelas, 2017;Hyland, Blaine, Dickenson, & Higgins, 2015). Once released into the environment, they possibly impact on the development of multi-resistant bacteria, with detrimental effects on human health as well as on the performance of naturally occurring nontarget species (Bártíkova et al., 2016;Jechalke, Heuer, Siemens, Amelung, & Smalla, 2014;Kumar, Lee, & Cho, 2012;Minden, Deloy, Volkert, Leonhardt, & Pufal, 2017). In plants, antibiotics can, among other effects, delay germination, reduce chlorophyll content and growth, and affect bioaccumulation (see summarizing tables in Bártíkova et al., 2016;Carvalho, Basto, Almeida, & Brix, 2014;Minden et al., 2017;Puckowski et al., 2016). It is notable that the effects of antibiotics on element contents of plants have received much less attention. To our knowledge, only one study has hitherto investigated effects of sulfadiazine on the C:N and K:Ca ratios of willow and maize plants, and found significantly lower C:N and K:Ca ratios in high antibiotic treatments (Michelini, Reichel, Werner, Ghisi, & Thiele-Bruhn, 2012).
However, this study used concentrations (200 μg/g soil) that are much higher than typically found in agricultural soils (0.006-500 μg/ kg soil, Thiele-Bruhn, 2003). The use of unnaturally high concentrations of antibiotics has recently been recognized as major drawback in relating results to in vivo situations (Bártíkova et al., 2016).
This study focuses on the effects of "in vivo concentrations" of veterinary antibiotics on element contents of plants. We tested how three different antibiotics (i.e., penicillin, tetracycline, and sulfadiazine) differing in their action modes affected element contents of four plant species, including crop (Brassica napus and Triticum aestivum) and noncrop (Capsella bursa-pastoris and Apera spica-venti) species. A previous study on antibiotic-effects on the performance of these plant species revealed significant effects on chlorophyll content, growth rates, and biomass allocation on the target organisms (Minden et al., 2017). Both crop species (B. napus and T. aestivum) belong to the most commonly grown crops worldwide (FAO, 2016;Leff, Ramankutty, & Foley, 2004) and are thus highly likely exposed to antibiotics due to fertilization of crop fields with slurry or manure. The noncrop species (C. bursa-pastoris and A. spica-venti) are commonly found along most crop field margins in Germany and are likely unintentionally exposed to antibiotic-charged manure applied to fields (Ellenberg & Leuschner, 2010). We applied concentrations of antibiotics as previously reported for grasslands (from now on referred to as in vivo concentrations, Thiele-Bruhn, 2003) to plants grown in greenhouses and measured macro-and micronutrients (N, P, K, C, Ca, S, Cu, Mg, Fe, Mn, Na) in fully developed plant individuals.
We investigated (a) whether in vivo concentrations of antibiotics generally affected element contents of plants and (b) whether plant responses differed between antibiotics, antibiotic concentrations and plant organs. Here, we expected roots to be most affected as shown by previous studies on antibiotics (Migliore, Rotini, Cerioli, Cozzolino, & Fiori, 2010;Pierattini, Francini, Raffaelli, & Sebastiani, 2016). We further tested whether scaling relationships between elements of different plant organs were either concurrent or discontinuous with the patterns of isometry and anisometry found by Kerkhoff et al. (2006) for plants grown under natural conditions.

| Selected plant species
The experiment was carried out with two crop species and two

| Selected antibiotics and their modes of action
The three antibiotics used in our study are penicillin G sodium salt (C 16 H 17 N 2 NaO 4 S), sulfadiazine (C 10 H 10 N 4 O 2 S), and tetracycline (C 22 H 24 N 2 O 8 ). They are the most commonly sold antibiotic compound classes used for food-producing animal species in Europe, with 37%, 23%, and 11% of sold antibiotics, respectively (European Medicines Agency, 2013; Tasho & Cho, 2016). As they are polar (with logKW < 3) they likely accumulate in plant tissue (Trapp & Eggen, 2013).
In general, both biodegradation and modes of action differ between the different types of antibiotics. Half-life for sulfadiazine is 50 days, whereas soil-stability for penicillin and tetracycline ranges from 40 days to 2 years, respectively (Christian et al., 2003;Hamscher, Pawelzick, Hoper, & Nau, 2005;Kumar, Gupta, Baidoo, Chander, & Rosen, 2005). Penicillin G (β-lactam antibiotic) inhibits the biosynthesis of peptidoglycan during microbial cell division and thus cell wall synthesis (Hammes, 1976;Miller, 2002). Sulfadiazine inhibits the growth of bacteria without destroying them (bacteriostasis) (Henry, 1944). Tetracycline is an anti-infective agent inhibiting protein synthesis by preventing the attachment of aminoacyl-t-RNA to the ribosomal acceptor (Chopra & Roberts, 2001
Converted to the amount of sand in the pots, treatments corresponded to 0.038, 0.19, and 0.38 μg/kg sand (see description of greenhouse experiment below). Antibiotics were ordered at Alfa Aesar (Karlsruhe, Germany). Antibiotic solutions were prepared by dissolving 1 mg of antibiotic in 1 L distilled water, before adding up 1 ml (5 and 10 ml) of removed solution to distilled water for a final volume of 1 L; pHs of all solutions were 5.5.
As all antibiotics used contain a nitrogen group (i.e., one molecule penicillin contains 7.8% N, tetracycline 6.3% N and sulfadiazine 22.4% N), we included one nitrogen (N-) treatment to differentiate between potential plant responses to antibiotics and/or to nitrogen provided by antibiotic degradation. We chose the highest amount of nitrogen provided by the antibiotics treatments (i.e., sulfadiazine treatment) as applied in the 10 μg/L treatment. Thus, 13.58 mg NaNO 3 were diluted in 1 L distilled water and 1 ml of this solution was further diluted with 1 L distilled water.
Macro-and micronutrients (N, P, K, Ca, S, Cu, Mg, Fe, B, Mn, Zn, Mo) were evenly applied to each experimental pot (5 ml solution/ week). Nitrogen was applied as NaNO 3 and phosphorus as NaH 2 PO 4 .
Composition of nutrient solutions followed Güsewell (2005), pH was adjusted to 6.

| Greenhouse experiment
Ten individuals per plant species were exposed to each treatment, summing up to 110 individuals per species and 440 individuals in total. Plants were raised from seeds in germination pots with germination soil (Gartenkrone, Germany). In June 2015, about 3 weeks after sowing, individual plants were planted in 400 ml pots filled with quartz sand (Vitakraft, Germany), B. napus was planted in 2-L pots. We used quartz sand instead of potting soil to guarantee a homogenous substrate for all treatments and thus prevent variation in soil-related factors (e.g., water-holding capacity) across pots. Also, antibiotics are organic compounds with a tendency to adsorb to soil particles, depending on soil pH, soil organic matter, and soil minerals (Tasho & Cho, 2016;Tolls, 2001). Using quartz sand, we restrained adsorption to soil organic matter, which can be strong (Thiele, 2000).
We mixed 25 ml of antibiotic and/or nitrogen solution with the sand before the seedlings were planted (125 ml for the 2-L pots).
This volume was equivalent to the quantity held back by quartz sand without draining. To avoid leaching of the antibiotics from pots, distilled water was filled into saucers only when needed. Nutrient solutions were provided once a week for 8 weeks in total. Control treatments received only distilled water and nutrients. Pots were randomly distributed in the greenhouse and shuffled once a week.
At the end of the experiment (i.e., after 8 weeks), plant individuals were harvested and separated into leaves, stems, and roots, dried at 70°C for 72 hr and weighed. Dead leaves were sorted and excluded from further analysis. Dried material of each organ of each harvested plant individual was ground in a planetary mill at 300-400 revolutions ("pulverisette 7"; Fritsch, Idar-Oberstein, Germany). For C:N analysis, each sample was further dried at 105°C for 4-5 hr.
Then, 2-3 mg of material were placed into tin tubes (0.1 mg precision balance CP 225 D; Sartorius, Goettingen, Germany) and analyzed using a CHNS Analyser Flash EA (Thermo Electron Corp., Waltham, MA, USA) following Allen (1989). All other elements (P, K, Ca, S, Cu, Mg, Fe, Mn, Na) were analyzed using optical emission spectrometry with inductively coupled plasma (ICP-OED: iCAP 6000 radial plasma view, Thermo Scientific), for which 8-10 mg ground material were digested with nitric acid and hydrogen peroxide. For every element, we used at least two spectral lines to identify interferences.
A solution with several internal standards was added to sample and calibration solutions for improving precision. Allocation of internal standard spectral lines to the analyte lines was chosen to match the product of ionization potential and energy of the spectral line. All values of plant elements refer to mg/g. For primary functions of elements see Supporting Information Appendix 1.
We first tested for significant differences between the nitrogen treatment and the control treatment, with the hypothesis that nitrogen addition in such small amounts should not have an effect on plant elements. Indeed, analyses yielded no differences, and data of the nitrogen treatment and the control treatment were thus pooled into one control treatment in subsequent analyses.
To test for effects of plant species identity, plant organs, type of antibiotics and their concentrations on the response variables (i.e., plant element content), multifactorial analyses of variance (ANOVA) were carried out. Factors were species (four levels: B. napus, C. bursa-pastoris, T. aestivum and A. spica-venti), plant organ (three levels: leaves, stems and roots), antibiotics (three levels: penicillin, sulfadiazine and tetracycline), and concentration (four levels: 0, 1 5, 10 μg/L) as well as their interactions. Level 0 of the concentration treatment refers to the pooled nitrogen and the control treatment (i.e., no application of antibiotics).
As our main research aim was to investigate whether in vivo concentrations of antibiotics generally affect element contents of plants, we subsequently performed one-way ANOVA between the control and antibiotic treatments (P1, P5, P10, S1, S5, S10, T1, T5, and T10, respectively) for each element separated for plant species and plant organ. Tukey's honest significant differences tests with false discovery rate correction was used to assess differences between individual groups (i.e., between control group and antibiotic treatments) (Benjamini & Hochberg, 1995;Noble, 2009). For all tests, element contents were transformed (log, boxcox) if necessary to meet the assumptions of parametric testing.
To describe bivariate relationships between element contents of different plant organs, we used standardized major axis regression (SMA). This analysis summarizes the relationship between two variables by minimizing the residuals in both variables (Kerkhoff et al., 2006;Warton, Wright, Falster, & Westoby, 2006), rather than predict one variable from the other (e.g., Y from X), which would be best described by ordinary least squares regression (Niklas, 2006).
To test for significant deviations from isometric scaling (slope ~1) between the element contents of the plant organ combinations, we To test our hypothesis that patterns of isometry and anisometry as described by Kerkhoff et al. (2006) may be disrupted by antibiotics, we, therefore, merged the plant species-specific datasets to one pooled dataset containing all four plant species and repeated the analyses.

| RE SULTS
Mean values for N, P, K, Ca, Mg, and S were highest in leaves, second highest in roots and lowest in stems (see Figures 1, 2,  . Asterisks indicate significant differences between control treatment and antibiotic treatment at p < 0.05, according to Tukey HSD test with false discovery rate correction F I G U R E 2 Effects of antibiotics on plant elements (mg/g) concentrations in stems of Brassica napus, Capsella bursapastoris, Triticum aestivum, and Apera spicaventi. Red bars show control treatments (C), antibiotics treatments were penicillin (P, yellow bars), sulfadiazine (S, green bars), and tetracycline (T, blue bars), in concentrations of 1, 5, and 10 μg/L. Values presented are the means of 10 replicates with the standard deviations shown in vertical bars (20 replicates for the control treatment). Asterisks indicate significant differences between control treatment and antibiotic treatment at p < 0.05, according to Tukey HSD test with false discovery rate correction F I G U R E 3 Effects of antibiotics on plant elements (mg/g) concentrations in roots of Brassica napus, Capsella bursapastoris, Triticum aestivum, and Apera spicaventi. Red bars show control treatments (C), antibiotics treatments were penicillin (P, yellow bars), sulfadiazine (S, green bars), and tetracycline (T, blue bars), in concentrations of 1, 5, and 10 μg/L. Values presented are the means of 10 replicates with the standard deviations shown in vertical bars (20 replicates for the control treatment). Asterisks indicate significant differences between control treatment and antibiotic treatment at p < 0.05, according to Tukey HSD test with false discovery rate correction TA B L E 1 F-values, degrees of freedom (df), and significance levels for multi-factor ANOVA analyses testing the effects of plant species (S), organ (O), antibiotic (A), and concentration (C) on elements in plant tissues (mg/g)  remaining plant elements, interactions between two or three factors yielded significant results (see Table 1). Overall, the factors species and organ had the strongest effects on plant elements, followed secondly by antibiotics and their concentrations. However, interactions between antibiotics and concentrations with species and plant organs affected all plant elements but Ca and S. Antibioticeffects differed between species (S × A, significant for N, P, C, Cu, Fe) and between plant organs (O × A, significant for K, Cu). Also, concentration-effects differed between species (S × C, significant for N, P, K, Cu, Mg, Fe, Mn, and Na) and between plant organs (O × C, significant for N, P, K, Cu, Mg, Na).

Pairwise comparisons between control and antibiotic treatments
showed that element contents in roots were most strongly affected by antibiotics, whereas stem element contents were weakly affected (Figures 1, 2, and 3). We found 50 comparisons between control and antibiotic treatments (Tukey HSD tests) to be significant for roots (corresponding to 13% of all tests for this plant organ), 20 for leaves (5% of all tests), and 13 for stems (3%), most of them in the penicillin treatments (Figures 1, 2, and 3). K was most responsive to the treatments, followed by N, C, S, and Cu. It is interesting that the values of element contents were mostly lower in the antibiotic treatments than in the control treatments for the two herb species (for N, K, Fe, and Na), whereas the two grass species showed reverse patterns: for P, K, Mg, and Na, element contents were higher in the antibiotics treatments than in the control treatments for T. aestivum and A. spica-venti, respectively (Figures 1, 2, and 3, and Supporting Information Appendix 3). Table 2 shows the results of SMA regression for the pooled dataset, in which all separate species-specific datasets were merged. We found significant correlations for element contents between organs for 92 of the 132 tests performed, of which most were detected between roots and stems. The slope of the regression as a measure of the extent of the increase or decrease of one element in two plant organs relative to each other indicates isometric or anisometric scaling relationships between these plant organs. Following the patterns derived from Kerkhoff et al. (2006), we expected the slopes between stems and leaves to be greater than 1 (α > 1, with stems on the y-axis and leaves on the x-axis), between roots and stems to be 1 (α ~ 1) and between roots and leaves to be smaller than 1 (α < 1, with roots on the y-axis and leaves on the x-axis). The data supported these expectations in 70% of all significant results for stems versus leaves. Significant slopes for roots versus stems were mainly <1 and for root versus leaves either ~1 or >1.
Among treatments, expected patterns were mostly found in control treatments (14 of 24 tested cases, indicated by ticks in Table 2), but less so in the antibiotic treatments (10 of 22 for tetracycline, 7 of 21 for sulfadiazine, and 6 of 25 for penicillin, see also examples in Figure 4).

| D ISCUSS I ON
To our knowledge, this study is the first to demonstrate effects of in vivo concentrations of antibiotics on plant organ element contents.
Our results show that antibiotics, even in these low concentrations, affect plant element content, with similar effects expected for realworld scenarios. We show significant species-specific responses as well as organ-specific responses to different antibiotics, with roots responding most strongly.
Our analyses showed that all elements responded to the antibiotic treatments, but with different response directions. Some element contents were lower in the antibiotic than in the control treatments, while the reverse was found for other elements. In general, plant element contents were most responsive to penicillin, followed by sulfadiazine and tetracycline, which was also found for other traits in the same four species and experimental setup (i.e., biomass allocation, growth rates, chlorophyll content and others, see Minden et al., 2017).
Moreover, responses to antibiotics differed between plant species: element contents in the two herb species B. napus and C. bursa-pastoris were mostly reduced by antibiotics, whereas contents were increased in the two grass species T. aestivum and A. spica-venti. Species-specific results have been shown by other studies on antibiotic-induced responses, for example, on germination Minden et al., 2017), postgerminative development (Migliore et al., 1997), root elongation (Pan & Chu, 2016), root and shoot lengths (Hillis et al., 2011), or bioaccumulation (Migliore, Brambilla, Cozzolino, & Gaudio, 1995).
It is interesting that organ-specific responses were unrelated to species, family, or functional group. Roots were, in general, most strongly affected by the treatments and stems were least affected.
To TA B L E 2 Slopes of standardized major axis (SMA) regression, confidence intervals, and correlation coefficients (r) for all combinations of plant organs (in the order of Y vs. X: stem vs. leaf, root vs. stem, and root vs. leaf) for each element across all species (pooled dataset) water uptake (Michelini et al., 2012;Piotrowicz-Cieslak, Adomas, Nalecz-Jawecki, & Michalczyk, 2010).

SMA of N
Plants regulate the uptake rates of elements to ensure element homeostasis (i.e., maintenance of constant body concentrations despite fluctuations in environmental resources, Bradshaw, Kautsky, & Kumblad, 2012;Cannon, 1929). Homeostatic regulation is needed to maintain element contents above certain thresholds under which growth is impaired, whereas excessive uptakes of elements can cause toxic effects (Güsewell, 2004). Antibiotics, once taken up by the root, may be transported to the aerial tissues (i.e., leaves, stems, flowers, fruits) of plants (Briggs, Bromilow, & Evans, 1982;Trapp & Mc Farlane, 1995), and may be stored, transformed or sequestered by plant metabolic processes ("green liver model," Burken, 2003;Sandermann, 2004 is to what extent our results were driven by antibiotics alone or by a combination of antibiotics and (impeded) microbial activity. As such, the results of the present study only reflect the responses of plant element contents to the antibiotic treatments, while we cannot make a distinction into direct (uptake and metabolization of the compound by the plant) and indirect (through microbial activity) effects of antibiotics. Kerkhoff et al. (2006) found patterns of isometric and anisometric scaling relationships between leaves, stems, and roots (and reproductive structures). The results of our pooled dataset of all four species supported their findings for stems versus leaves (Y vs.

SMA of S
X). However, for the remaining combinations (roots vs. stems and root vs. leaves), their patterns could only partly be reproduced.
Moreover, expected isometric and anisometric scaling relationships were most often found for the control treatments, but less so for the antibiotic treatments, with penicillin yielding the lowest number of concordant patterns. This is well in line with our results revealing strongest effects on tissue element concentrations for penicillin and weakest effects for tetracycline.
A similar analysis on how stressors may affect scaling relation- ships, yet with field data from salt marshes, has been conducted by Minden and Kleyer (2014). They concluded that "structural" organs, like stems and roots, were less homeostatic (i.e., they exerted high elemental fluctuations due to environmental constraints) than "metabolic" organs like leaves and diaspores. However, they identified stems as the organs with lowest homeostasis in response to nutrient availability and salt stress. Our results indicate that under antibiotic stress, roots exhibit a lower homeostasis than under natural conditions. A low homeostatic element composition in one organ also implies effects on other plant organs, because the overall performance of a plant depends on the interplay between its organs and their specific functions (see Kleyer & Minden, 2015). In a consequent manner, low homeostatic responses in one or more plant organs may affect the performance of the whole plant, which may further scale up to the community level with consequences for community composition and other trophic levels.
In summary, the results from our study show that antibiotics in concentrations as found in agricultural landscapes can affect element contents of plants, particularly in roots. In roots, they may lead to low-element homeostasis, altering the scaling between roots and other plant organs, which may affect metabolic processes and ultimately the performance of a plant.

ACK N OWLED G M ENTS
The authors thank Andrea Deloy and Anna Martina Volkert for their support with greenhouse work and Daniela Meißner and Eleonore Gründken for assistance in the laboratory. We are grateful for the comments provided by two anonymous reviewers, which greatly helped to improve our manuscript. Funding was provided by the Deutsche Forschungsgemeinschaft (DFG project: MI 1365/3-1).

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
VM, GP, and SDL designed research; VM performed research; VM and BS analyzed samples; all authors contributed critically to the draft and gave final approval for publication.