Sphagnum growth under N saturation: interactive effects of water level and P or K fertilization

(cid:129) Sphagnum biomass is a promising material that could be used as a substitute for peat in growing media and can be sustainably produced by converting existing drainage-based peatland agriculture into wet, climate-friendly agriculture (paludiculture). Our study focuses on yield maximization of Sphagnum as a crop.


INTRODUCTION
Sphagnum biomass is an important renewable raw material that can be used in various products (Pouliot et al. 2015;Glatzel & Rochefort 2017). It can substitute for fossil peat, especially for slightly humified Sphagnum peat (Emmel 2008;Reinikainen et al. 2012;Blievernicht et al. 2013;Jobin et al. 2014), which, with an annual worldwide consumption of 30 million m 3 , is the major constituent of growing media used in professional horticulture (Schmilewski 2017). Its cultivation on rewetted peatlands ('Sphagnum farming') contributes to reducing greenhouse gas emissions from formerly drained agricultural land (Beyer & H€ oper 2015;Wichtmann et al. 2016;G€ unther et al. 2017). Replacing peat with Sphagnum biomass, however, requires a substantial expansion of the area under cultivation and of the yield per hectare (Gaudig et al. 2014Wichmann et al. 2017). Maximizing yield implies that conditions for optimal Sphagnum growth should be identified.
Sphagnum growth rate is, next to climate factors, determined by water and nutrient availability as well as the Sphagnum species (Gunnarsson 2005). Under natural conditions, atmospheric water supply in NW Germany used to provide sufficient water to cover water losses, in particular through evapotranspiration in summer. This is, however, no longer the case. In the present situation of higher atmospheric water demand and larger seepage losses as a result of extensive drainage in the surrounding land, Sphagnum farming now requires an additional water supply (Brust et al. 2018) to keep the photosynthetically most active apical capitulum continuously moist (cf. Robroek et al. 2007). Several studies have shown that the growth rate of most Sphagnum species is highest at water tables just below the capitula, independent of the species (Clymo & Reddaway 1971;Hayward & Clymo 1983;Campeau & Rochefort 1996;Robroek et al. 2009;Brown et al. 2017).
In NW Germany, with a total atmospheric N deposition of ca. 38 kgÁha À1 Áyear À1 (Gauger et al. 2002), Sphagnum growth is not N-limited, and the N pool of the Sphagnum layer is saturated (cf. Malmer 1990;Lamers et al. 2000;Berendse et al. 2001;Bragazza et al. 2004). Additional N input would not result in extra Sphagnum growth and could even negatively affect growth, both directly (e.g. by lower photosynthesis, increased metabolic costs) and indirectly (by vascular plants increasingly competing with Sphagnum for light and water; Berendse et al. 2001;Tomassen et al. 2004;Limpens et al. 2011;Fritz et al. 2014). Under high N loads Sphagnum growth may furthermore become limited by phosphorus (P) (Aerts et al. 1992;Verhoeven et al. 1996;Lund et al. 2009) and potassium (K) (Bragazza et al. 2004). The addition of P may then substantially increase Sphagnum biomass production Fritz et al. 2012), but this effect disappears with insufficient water availability (Aerts et al. 2001;Limpens et al. 2004;Fritz et al. 2012). High Sphagnum biomass production was indeed observed in a Sphagnum farming field experiment in NW Germany under high N loads, balanced supply of P and K in irrigation water, and year-round water tables just below the moss surface (Temmink et al. 2017). However, the water table varied up to 20 cm over time and around 10 cm in space because of differences in micro-relief (cf. Brust et al. 2018).
In this study we address the relationship between nutrient supply, high water level regimes and growth of different Sphagnum species for the first time in a glasshouse factorial experiment. We hypothesize that under N-saturated conditions, the highest Sphagnum growth rates can be achieved by combining a constantly high water table with extra P and K fertilization.

MATERIAL AND METHODS
The experiment was set up in a full factorial design with three replicates and repeated twice with the same combination of water regime and fertilization level. A total of 216 moss containers (4 species 9 3 water regimes 9 3 fertilization levels 9 3 replicates 9 2 repetitions) were placed in 18 (2 repetitions 9 9 combinations of water regime and fertilization level) boxes (4 9 60 9 23 cm) filled with culture medium after Rudolph et al. (1988) (Table 1). Each box contained 12 containers: three of each species (with the exception of S. palustre with lower availability) distributed randomly within the box. To ensure similar conditions within the boxes, 'gaps' due to the lack of S. palustre were filled using containers with S. papillosum (same section).

Sphagnum species
Pure patches (12 9 12 cm and 10-cm deep) of Sphagnum papillosum, S. palustre, S. fimbriatum and S. fallax were collected from natural lawns in Lower Saxony (Esterweger Dose, NW Germany) and placed in containers (12 9 12 9 12 cm) with a perforated base, within the boxes filled with a culture medium.

Water regimes
Three water regimes were applied by adjusting water levels in the boxes each week: (i) water level rising with moss growth and remaining 2 cm below the top of the capitulum ('rising'), (ii) water level alternating between 2 cm (1 week duration) and 8 cm (3 weeks duration) below the top of the capitulum ('fluctuating'), (iii) water level starting 2 cm below the top of the capitulum and remaining at the same absolute level in spite of moss growth ('static').

Cultivation conditions
The mosses were cultivated in a glasshouse with a light regime of 12-h light (mostly sunlight, but at light flux densities < 15 klx, supplemented with a sodium vapour lamp -Philips Son-T Agro 400 Wof 80 µmolÁm À2 Ás À1 ) and 12-h dark. Air temperatures in the glasshouse were 18-21°C during the day and 12°C at night, without active cooling. Temperatures at the moss surface never exceeded 35°C and were thus not detrimental to the mosses (cf. . After 2 weeks of adaptation, the experiment ran from September 2006 to June 2007 (280 days). Sprinkler irrigation (~0.25 mm demineralized waterÁm À2 ) was applied twice a day (06:00 and 18:00 h) for 1 min, simulating morning dew and rain (Rudolph 1963). Different growth rates between containers within a box were corrected weekly by adjusting the height position of the containers. After 54 and 146 days, all mosses of the 'rising' and 'fluctuating' treatments were cut back from the base to a length of 10 cm to maintain similar water levels within a box for all containers. In the 'static' treatment, container walls were regularly elongated to avoid interaction between the growing mosses. Table 1. Nutrient availability during the experiment (280 days) -calculated from nutrient concentrations in the boxes at the beginning of the experiment ('culture medium', pH 5.8) and the nutrients added by fertilization (fertilization solution: pH 8.7-9.0, weakly buffered) -depending on fertilization treatment compared to mean concentration of the culture medium in the boxes at the end of the experiment (in kgÁha À1 AE SEM, n = 6 for each fertilization treatment). N availability was the same for all treatments, with N at 6.67 kgÁha À1 by culture medium + 28.98 kgÁha À1 by adding fertilizer = 35.65 kgÁha À1 total. C = control, P5 = five-fold addition of phosphorus, 2P2K = double addition of both phosphorus and potassium.

Growth measurements
At the beginning of the experiment, five moss shoots per container were marked with nylon zip ties (width 2 mm, length 100 mm) fixed between the capitulum and the subjacent branches (cf. Overbeck & Happach 1957;Clymo 1970). At the end of the experiment (day 280) all mosses were cut at the level of the zip ties and at 1 cm below the surface to separate the capitula (0-1 cm). Length increase was measured, and biomass weight of each marked moss shoot and each container determined after drying for at least 48 h at 60°C. Annual biomass productivity was calculated by extrapolating the subcapitulum weight (dry mass) of each entire container to hectares and year, assuming that the biomass of the capitula had not changed since the end of the initial adaptation period. Weight per length unit was calculated for each marked moss shoot to characterize compactness.
Fruiting bodies of the fungus Sphagnurus paluster were removed from 28 of the 216 containers during the experiment to prevent further distribution of the fungus. The cover (% of the container area) of Sphagnum tissue with necrosis (bleached capitula) and with algal infestation (dark green capitula) was estimated at the end of the experiment.

Nutrient measurements
The biomass of all capitula per container was dried and milled in a centrifugal ball mill (Pulverisette 14, Fritsch Idar-Oberstein; for 1-2 min at RCF: 15,580 g), and total N concentration and C/N ratio determined with a dry-combustion C/N analyser (CHNOS element analyser; Vario EL III, Elementar Analysensysteme, Hanau, Germany). After dry ashing (in a muffle furnace at 550°C for 4 h), the ash was dissolved in 10% H 2 SO 4 (Kalra 1998) and the solution treated with an acidic molybdate solution containing ascorbic acid (modified molybdenum blue method; Temminghoff 2004) to measure total P using a UV/ Visible spectrophotometer (Cecil CE 1021, 890 nm wavelength). Potassium (K) was determined with an atomic absorption flame spectrometer (CD-ContrAA 300, analytic Jena) directly after microwave digestion (START 1500, MLS Enterprises). The K concentration of the water in the boxes was determined as described for the biomass samples but without digestion. Orthophosphate (ortho-P) in the water was measured after filtration (cellulose acetate filter with 0.45-µm pore size) using the modified molybdenum blue method (Temminghoff 2004), and ammonium (NH 4 + ) was measured spectrophotometrically using the salicylate method (Krom 1980).

Data analysis
We analysed the effects and possible interactions of the treatments, Sphagnum species, water regime and fertilization level, on Sphagnum dry mass productivity and length increase, moss compactness, N, P and K concentrations and N/P and N/K quotients ratios in the Sphagnum capitula, and cover of necrosis and algae (dependent variable). As we had different sample sizes (number of replicates) and no homogeneity among the datasets, assumptions necessary for applying linear regression models (including ANOVA models) were not met (Zuur et al. 2009). To accommodate possible spatial correlation of Sphagnum containers of the same species in one box, we thus applied linear mixed effect models with fixed and random components (Pinheiro et al. 2009;Zuur et al. 2009). We also applied linear mixed effect models to compare the results of dry mass productivity (tÁha À1 Áyear À1 ), rate of increase in length (cmÁyear À1 ), compactness (mgÁcm À1 ) and cover of necrosis and algae (% per container) for each Sphagnum species (S. fallax, S. fimbriatum, S. palustre and S. papillosum) with regard to the treatments water regime and fertilization level.
Restricted maximum likelihood estimation (REML) was used to calculate estimates of coefficients for the models (Zuur et al. 2009). To identify the optimal model, we used the Akaike information criterion (AIC), which measures goodness-of-fit and model complexity (the lower the AIC value, the better the model).
Furthermore, we measured the strength and direction of association between dry mass productivity and necrosis, algae and N, P and K concentrations in the Sphagnum capitula using Pearson product-moment ('standard') or Spearman rank-order correlation, depending on whether the data were normally or not normally distributed, respectively. Correlation between Sphagnum dry mass productivity and N/P ratio was analysed with a generalized additive model with integrated smoothness estimation (Wood 2006).
Data exploration, computation and figure design were done with the software R (R Development Core Team 2009) and the packages' 'nlme' (Pinheiro et al. 2009), 'mgcv' (Wood 2006) and 'stats' (R Development Core Team 2009).
Length increases and biomass productivity of all Sphagnum species were highest with the water table staying constantly 2 cm below the capitulum (treatment 'rising'; Table 2). Moss growth decreased with lower water tables, even if lowering was only periodic and only a few centimetres. As long as high water tables (2 cm below capitulum) occurred periodically, S. palustre grew better than with a water level not rising with moss growth (treatment 'static', i.e. water table sinking relative to the moss growth; P = 0.027). At the end of the experiment, the water level in the 'static' water level treatment was: S. fallax 1-9 cm (mean 5.6 cm), S. fimbriatum and S. palustre 4.0-8.5 cm (mean 6.3 cm) and S. papillosum 3.0-7.5 cm (mean 5.3) cm below the capitulum. The 'static' water level led to the lowest biomass and length values for all species, except biomass productivity of S. fallax, which was similar to that with a fluctuating water table (Fig. 1, Table 2).
Fertilization with P or with P and K had no effect on Sphagnum growth (Table 2).

Compactness
Compactness, i.e. dry mass per unit moss length, determines water-holding capacity and capillarity (cf. Titus & Wagner 1984). Compactness was used as a proxy for the suitability of Sphagnum biomass as a raw material for growing media (cf. Jacobs et al. 2009).
Sphagnum papillosum and S. palustre were significantly more compact than S. fallax and S. fimbriatum ( Fig. 1; Table 2). A 'static' water level resulted in more compact growth of all species, but this was only significant for S. palustre ( Fig. 1; Table 2). Fertilization had no significant influence on the morphological characteristics of these peat mosses.

Nutrient concentrations in Sphagnum capitula
The N concentrations in all Sphagnum species were similar (mean N 14.4 mgÁg À1 DW). Generally, biomass productivity decreased with increasing N concentration in the capitulum (Pearson correlation: r = À0.23, n = 190, P ≤ 0.01). Sphagnum papillosum had the lowest biomass production with highest N values (max. 26.2 mgÁg À1 DW).
The P concentrations ranged between 0.63 and 2.36 mgÁg À1 DW (mean 1.3 mgÁg À1 DW), K concentrations were between 2.9 and 11.9 mgÁg À1 DW (mean 6.0 mgÁg À1 DW). While lowest P concentrations were in S. papillosum (mean 1.1 mgÁg À1 DW), the highest K concentrations were in S. palustre and S. fallax with 'rising' water level (Table 3). Fertilization did not influence N, P or K concentrations in the capitula nor the growth (Tables 2 and 3). Both P and K concentrations were lowest at the 'static' water level (Table 2). Biomass productivity decreased with increasing N/P values (Table 2; Fig. 3), with highest N/P ratio reached in S. papillosum (N/P = 7-24).
The N/K ratio ranged between 1.1 and 5.3 but had no relationship to biomass productivity. On the other hand, N/K values were significantly lowest in the 'rising' water table, except for S. fallax (Table 3).

Necrosis and algae
Algae (leading to dark green capitula) were recorded in 38% of the containers. Also, in 38% of the containers, more than 5% of the Sphagnum plants suffered from necrosis, probably because of fungal infection. With higher water tables there were fewer necrosis and algal infections (Fig. 2). Containers with S. fimbriatum experienced the highest level of necrosis per container (up to 92%), but proportionally more containers of S. palustre were affected. Sphagum fallax experienced the highest level of algal infestations (up to 55% of the moss; Fig. 2).
We excluded six containers with S. fallax from data analysis because these mosses collapsed and lost their structure, making further measurements impossible. The reasons for the die-off are unclear. There was no link between die-off and treatment.
Biomass productivity values as a function of cover of necrosis (% per container) were wide ranging (Fig. 4), but decreased Fig. 1. Dry mass productivity (tÁha À1 Áyear À1 ), length increase (cmÁyear À1 ) and compactness (mgÁcm À1 ) of Sphagnum fallax, S. fimbriatum, S. palustre and S. papillosum as a function of water regime (see text for details), with levels of significance for comparison to 'rising' water level within one species (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001) and total number of measurements (N compactness is similar to N length increase). The 'box and whisker' graphs show the median (bold line), the upper and lower quartiles (which include 50% of the data and create the box), whiskers representing the lowest value within 1.5 IQR (=interquartile range) of the lower quartile and the highest value within 1.5 IQR of the upper quartile, and outliers (o, i.e. values outside these ranges).
with increasing percentage of necrosis, particularly in the 'rising' water table, were less distinct in the 'fluctuating' water table, whereas in the 'static' water level biomass productivity was independent of necrosis ( Fig. 4). We observed an association between decreasing Sphagnum dry mass and an increase in algae (Spearman correlation: r s = À0.42, n = 191, P ≤ 0.001). Fertilization had no significant effect on necrosis or algae, but algal occurrence was higher in the 5P treatment. Furthermore, there was no relationship between nutrient concentration (N, P, K) in the moss capitula and percentage of necrosis or algae.

Water level
Our results show that constantly high water tables, i.e. continuously rising with the growing moss, lead to the highest growth rates for all four Sphagnum species. As soon as the relative water table falls by only a few centimetres, Sphagnum growth is significantly hampered. The 'static' water level, which sank relative to the up-growing moss, apparently hampered Sphagnum growth more than an alternating water level (treatment 'fluctuating'). This concurs with the results of Robroek et al. (2007), who found a lower capitulum water content and a consequent growth reduction in lawn species growing with a water table of 15 cm compared to a water table 5 cm below the top of the capitulum. In contrast, Breeuwer et al. (2009) found the productivity of the lawn species Sphagnum magellanicum increased with summer water table fluctuations of between 7 and 23 cm in comparison to somewhat wetter conditions (water table 3-15 cm below moss surface), and they attributed this to a competitive advantage of S. magellanicum in the drier conditions over the co-occurring S. cuspidatum. Without competition, growth of lawn species is generally highest at high water levels (Hayward & Clymo 1983;Grosvernier et al. 1997;Johnson 1998;Stokes et al. 1999).
Despite their more dense and compact growth form, our mosses obviously could not compensate for lower water levels through more effective capillarity (cf. . The water content of the capitulum is a good indicator of whether water supply is sufficient for optimal CO 2 assimilation (cf. Robroek et al. 2009). We did not measure capitulum water content in our study, but several other studies have found a rapid decrease when lowering water levels by only a few centimetres Robroek et al. 2009;Strack & Price 2009). On the other hand, even small amounts of precipitation (0.5-1.0 mm) may rewet the capitulum sufficiently to reduce the negative effect of low water levels Strack & Price 2009;Nijp et al. 2014;Krebs et al. 2016). As we sprinkled water on our mosses twice a day, with 0.5 mm per day, and each week replenished water loss in the boxes, capitulum water content will only have varied very slightly. A high water level not only leads to optimal water supply to the capitulum, but also to an improved nutrient supply (cf. , higher vitality (indicated by less necrosis and algal infestation) and a looser growth form, allowing light to penetrate deeper into the Sphagnum lawn, resulting in an increased active assimilation area (Sliva 1997;cf. Robroek et al. 2009). Continuous optimal growth requires a high water table that continuously rises with the growing moss. A water drawdown would result in a growth reduction orin the case of low stem densityeven death of the peat mosses (Fritz et al. 2012).

Fertilization
Since Sphagnum growth is not N-limited at atmospheric deposition rates exceeding 18 kgÁha À1 Áyear À1 (Malmer 1990;Aerts et al. 1992;Verhoeven et al. 1996Lamers et al. 2000 and our control (38 kgÁha À1 Áyear À1 ; Gauger et al. 2002; Table 1) far exceeded this value, we refrained from testing additional N fertilization. The mean Sphagnum tissue N concentration of 14.4 mgÁg À1 DW found at the end of our experiment, indeed confirms the prevalence of N-saturated conditions (Lamers et al. 2000).
Over the entire dataset there was a negative correlation between N tissue concentration and biomass productivity, but this effect was most distinct in S. papillosum (r = À0.53, P ≤ 0.01). Its growth strongly decreased at capitulum N concentrations >20 (max. 26.2) mgÁg À1 DW, resulting in the lowest biomass productivity values measured in this study. For the other species, N tissue concentration had no effect on growth (cf. Limpens & Berendse 2003). Berendse et al. (2001) proposed a maximum N concentration in Sphagnum tissues of 20 mgÁg À1 DW, which was exceeded in both the study of Breeuwer et al. (2009) and in our study. According to van der Heijden et al. (2000), a capitulum N concentration in S. fallax of 15 mgÁg À1 DW indicates N pollution stress in bogs. However, our study shows that even a maximum N value of 20.5 mgÁg À1 DW has no negative effect on growth of S. fallax. Table 3. Mean nitrogen, phosphorus and potassium concentrations (mgÁg À1 dry mass AE SEM) and N/K quotient in capitula of different Sphagnum species as a function of water regime and fertilization level. Similarly, Granath et al. (2009) found no detrimental effects on the photosynthetic apparatus at N tissue concentrations up to 20 mgÁg À1 DW in S. balticum (a species of the Cuspidata section, like S. fallax). Bragazza et al. (2005) suggested that Sphagnum plants in polluted regions have a metabolic adaptation (with lower rates of N absorption) to high N supply, which was confirmed by Fritz et al. (2014). This ability seems, however, to be differently developed between species, as also suggested in Fig. 3. In our study (as in Temmink et al. 2017 with N concentrations >18 mgÁg À1 DM), a toxic effect (growth reduction) of a high N tissue concentration >20 mgÁg À1 DM was observed only in S. papillosum. Chiwa et al. (2016) found that an S. capillifolium lawn can filter wet N deposition of up to 32 kg NÁha À1 Áyear À1 for least a decade, leading to N concentrations in the capitula of around 14 mgÁg À1 DM. In our study, P fertilization had no significant effect on Sphagnum growth, indicating the absence of P limitation. This is corroborated by the failing correlation between P concentration in the capitula (mean 1.3 up to 2.4 mgÁg À1 DW) and fertilization (cf. Li et al. 1993, Table 2), as well as N/P ratio <30 (cf. Bragazza et al. 2004). In other studies, with low P concentrations in the pore water, similarly high P concentrations in the mosses were found only after P fertilization (Chiwa et al. 2018, Limpens et al. 2003bLimpens et al. 2004;Limpens & Heijmans 2008;Fritz et al. 2012). As in our study, Sphagnum did not respond significantly to additional P (Li et al. 1993;Limpens et al. 2004) at sites with similar water P concentrations as in our experiment (0.23 mgÁl À1 ) and in natural bogs in Lower Saxony (Bertram 1988;L€ utt 1992). Only at 'rising' water level (remaining 2 cm below the capitulum) did P fertilization (5P) slightly (but not significantly) increased growth of Sphagnum palustre, S. fimbriatum and S. fallax in our study. The P concentrations in the culture medium were similar at the end of the experiment irrespective of the treatment (see Table 1). This can be explained by P fixation into forms that are unavailable to the plants, as only orthophosphate was determined in the solutions.
The N/P ratios in the Sphagnum capitula at the end of the experiment ranged between 7.2 and 23.6, with a mean value of 11.3, indicating optimal nutrient supply (cf. Aerts et al. 1992). With increasing N/P ratio, biomass productivity significantly decreased (Fig. 3). Nevertheless, no P   3. Correlation between dry mass productivity (tÁha À1 Áyear À1 ) and N/P quotient for different Sphagnum species; grey: N/P quotient with optimal Sphagnum growth (after Aerts et al. 1992). Model results (generalized additive model), N = 191: estimated degrees of freedom for the smoother = 3.23, explained deviance (R 2 ) = 30.5%, variance of the residuals = 1.25, P < 0.001. The curve is estimated using LOESS smoother and point-wise 95% confidence bands (dotted lines) (Zuur et al. 2009). limitation can be found in our study (see above). According to Aerts et al. (1992), N/P ratios below ten indicate N limitation, which would apply to about 40% of our values. As all mosses received the same (high) N amount, P fertilization had no influence, and Sphagnum species did not differ in N/P ratios, the suggestion of N limitation is implausible.
Compared to studies in natural habitats with 2.5-5.2 mg KÁg À1 DW (Bragazza et al. 2004;Fritz et al. 2012), the K concentration in the capitula in our study (mean 4.7-9.6 gÁg À1 DW; Table 3) was higher, but similar to a Sphagnum farming site (irrigated with eutrophic water) with 4.5-10.3 g KÁg À1 DW (Temmink et al. 2017). Bragazza et al. (2004) suggest K limitation at N/K ratios above 3.3, which was only found in a single case in our study (N/K 1.1-5.3, mean 2.6). However, K fertilization did not lead to either significantly higher biomass productivity nor higher K concentrations or lower N/K ratios in the moss capitula. Similarly, the K concentration in the culture medium at the end of the experiment was independent of the fertilization treatment. These facts indicate K-saturated conditions.
Higher N, P and K concentrations in the capitula at constantly high water levels ('rising' treatment) may result from increased nutrient uptake by the larger moss surface with permanent water contact (cf. . The values for N, P and K are known to be more concentrated in the Sphagnum capitula (Malmer 1988), but under nutrientsaturated conditions their accumulation in stems cannot be ruled out (cf. Chiwa et al. 2018); unfortunately, this was not measured in our study. Li et al. (1993), however, we did not find changes in P concentration either in the capitula (0-1 cm) or in the stem (1-4 cm) of S. papillosum at different P fertilization levels. Nevertheless, nutrients will have been removed from the rather closed box system in our study through the cutting and removal of the basal parts (see Methods).
Both the N/P and N/K ratios in our study indicate that conclusions from studies of natural systems cannot simply be transferred to systems with high nutrient loads and high nutrient concentrations in the moss tissue.
We used the culture medium of Rudolph et al. (1988), which is optimized for Sphagnum growth and was apparently sufficient and had a favourable stoichiometry, thus no stimulating effect of additional P and K fertilization on Sphagnum growth was found. In practice, even larger amounts of nutrients, in particular N, P and K, are supplied to the moss layer when the irrigation water of the Sphagnum farming site is obtained from the surrounding fertilized agricultural areas (Krebs et al. 2012;Temmink et al. 2017). Nutrient-rich conditions (as in our study) are representative for extensive areas of Western and Central Europe. Our study demonstrates that Sphagnum grows well under nutrient-rich conditions, as long as an optimal water supply is guaranteed. These results might also be useful for bog restorations. The long-term effects of such site conditions, e.g. on Sphagnum growth and species composition, still have to be investigated.
In contrast to the results of Fritz et al. (2012) for a nutrientpoor site, we did not find changes in Sphagnum morphology as a result of fertilization since N, P and K supply in our study were apparently sufficient for Sphagnum growth, including in the control.

Sphagnum species
Of the four studied Sphagnum species (S. palustre, S. papillosum, S. fimbriatum, S. fallax), S. fallax had the highest productivity, which corresponds to results from a global meta-analysis (Gunnarsson 2005). Both high water level and an adequate nutrient supply promote growth of all four tested species, but the minerotrophic species S. fallax profits most (cf. Lee & Studholme 1992;Twenh€ oven 1992;Limpens et al. 2003b). On the other hand, S. fallax decomposes faster than S. papillosum (Limpens & Berendse 2003) and has a lower water-holding capacity (Overbeck & Happach 1957), which might make this species less suitable for use in horticultural substrates, at least for some applications (Emmel & Kennett 2007).

Necrosis and algae
Necrotic diseases of peat mosses are often caused by pathogenic fungi, such as Sphagnurus paluster (syn. Lyophyllum palustris, Tephrocybe palustris) (Redhead 1981;Untiedt & M€ uller 1985;Limpens et al. 2003a); this parasitic basidiomycete is only found on Sphagnum (Untiedt & M€ uller 1985). Although no molecular identification of the fungal mycelium was conducted, the typical pattern of damage and sporocarps in our study indicate the occurrence of Sphagnurus paluster.
In contrast to Limpens et al. (2003a), our study found a negative effect of infection on biomass production of Sphagnum (Fig. 4). This contradiction may be explained by the increased fungal biomass being included in the biomass values of Limpens et al. (2003a) and by the intensity of necrosis being lower at high water levels (Fig. 2). Similarly, in contrast to Limpens et al. (2003a), we did not find any relationship between necrosis or algae and fertilization treatment or N/P ratio. The N/P ratio in our study, however, did not exceed 25, i.e. Fig. 4. Relation between dry mass productivity (tÁha À1 Áyear À1 ) and necrosis (% cover per container) for different water regimes with regression lines, including all species. Overall model with dry mass productivity (dependent variable) and necrosis (explanatory variable) and its interaction with the different water regime model results (linear mixed effect model), N = 192: degrees of freedom = 188, F-value = 3.58, P < 0.05, Rho (induced correlation structure of containers of the same moss species within a box) = 0.53. remained in the range where Limpens et al. (2003a) also failed to see any correlation. Our results indicate that Sphagnum vitality and growth rate are stimulated by high water levels, where they are less vulnerable to fungal or algal infection despite high nutrient loads.
While Sphagnurus paluster infects only small areas in nature, it often kills most peat mosses in glasshouses where there is a favourable environment for rapid fungal dispersal (Untiedt & M€ uller 1985;Landry et al. 2011). Since Sphagnum farming sites on rewetted bogs are also artificial systems, the risk potential for diseases from fungi and algae must be assessed. Effective measures to limit Sphagnurus paluster without affecting Sphagnum are the fungicide Myclobutanil (Landry et al. 2011) or Trichoderma virens as an antagonist (Irrgang et al. 2012), which have only been tested and might only be applicable in glasshouse cultivation.