Effects of tropospheric ozone and elevated nitrogen input on the temperate grassland forbs Leontodon hispidus and Succisa pratensis

Atmospheric ozone (O 3 ) and nitrogen (N) pollution have increased since pre-industrial times and pose a threat to natural vegetation. The implications of these pollutants for the perennial temperate grassland species Leontodon hispidus (Rough Hawkbit) and Succisa pratensis (Devil ’ s-bit Scabious) are largely unknown. Both species are important for pollinators and Succisa pratensis is the host plant for the threatened marsh fritillary butter ﬂ y ( Euphydryas aurinia ). We examine growth and physiological responses (leaf cover, leaf litter, ﬂ owering, chlo- rophyll index [ Leontodon hispidus and Succisa pratensis ]; photosynthesis and stomatal conductance [ Succisa pratensis ]) using an outdoor Free Air Ozone Enrichment system. Plants were exposed to Low, Medium and High ozone treatments over three growing seasons (treatment means: 24, 40 and 57 ppb, respectively), with and without the addition of nitrogen (40 kg ha (cid:1) 1 yr (cid:1) 1 ) during the ﬁ rst year. Decreases in leaf cover (p < 0.001) and chlorophyll index (p < 0.01) were observed with increased O 3 for Leontodon hispidus . The addition of N resulted in a higher chlorophyll index only at the uppermost O 3 level and also led to an overall increase in litter production of 6%. However, a stronger effect of both O 3 and N treatments was observed with Succisa pratensis. Litter production increased with increasing O 3 (p < 0.001) and an overall rise of 31% was recorded with added N (p < 0.05). However, O 3 had the biggest impact on Succisa pratensis foliage leading to more damaged leaves (p < 0.05). During summer resources were prioritised to new leaves, maintaining stomatal conductance and photosynthesis rates. However, this was not sustained during autumn and accelerated senescence occurred with higher ozone, and rates declined faster with added nitrogen (p < 0.05). Elevated O 3 also reduced Succisa pratensis ﬂ owering (p < 0.01). These effects have implications for inter- and intra-speci ﬁ c competition, seed establishment, nutrient cycling, as well as the provision of general pollinator resources with speci ﬁ c issues for butter ﬂ y larvae. Results highlight the need for concerted action to reduce pre-cursor ozone emissions to go alongside habitat management efforts to protect biodiversity.


Introduction
At ground level, ozone (O 3 ) is a greenhouse gas and secondary air pollutant. It is formed from solar radiation-driven chemical reactions between pre-cursor gases, including carbon monoxide (CO), nitrogen oxides (NO x ), methane (CH 4 ) and non-methane volatile organic compounds (Monks et al., 2015;Royal Society, 2008;Simpson et al., 2014). Concentrations have increased since pre-industrial times by around 40% and O 3 pollution continues to be a threat globally to human health, natural vegetation and crops (Cooper et al., 2014;Fleming et al., 2018;Mills et al., 2018aMills et al., , 2018bRonan et al., 2020). Whilst O 3 might affect plant community composition (Payne et al., 2011), little is known about its impact on biodiversity (Fuhrer et al., 2016). Atmospheric deposition of reactive nitrogen (N) is a major driver of biodiversity change, especially for nitrogen limited ecosystems where N deposition is a problem for many species (Sala et al., 2000;Bobbink et al., 2010). Dependent on climatic conditions, both O 3 and N can be transported distances away from urban and industrialised zones leading to pollution issues in rural areas .
In plants O 3 is predominantly taken up through leaf stomatal pores, where it reacts with the apoplastic fluid inside the leaf forming reactive oxygen species. In O 3 sensitive plants this can cause changes in cell structure and metabolism, and trigger cellular response defence mechanisms causing early senescence and the repression of growth and seed production (Ainsworth, 2017;Bergmann et al., 2017). Natural and semi-natural grasslands are a cause for conservation concern, being important globally for maintaining biodiversity and the provision of ecosystem services (Habel et al., 2013;Carbutt et al., 2017), and both O 3 and N sensitive taxa can be found as part of these ecosystems (Hayes et al., 2007;Stevens et al., 2010;Payne et al., 2013;Bergmann et al., 2017).
Recent (2013e2017) summer averages of O 3 concentrations across a range of European temperate zones have varied between 30 and 100 ppb (European Environment Agency (EEA), 2017). In grassland communities higher levels of O 3 have been found to result in visible leaf injury, increases and decreases in biomass (Bungener et al., 1999;Bergmann et al., 2017), changed timing of plant life-cycles (Hayes et al., 2011), and reduced pasture feed/forage quality (Gonz alez-Fern andez et al., 2008;Hayes et al., 2016), as well as impacting on species composition (Ashmore et al., 1995;Wedlich et al., 2012) and stomatal functioning (Mills et al., 2009;Wilkinson and Davies 2009). Whilst increased N can result in increased growth and biomass in some species, it can also reduce plant coverage and be harmful to species adapted to low N conditions (Bobbink et al., 1998;Payne et al., 2013). This can lead to changes in grassland species composition and richness (Stevens et al., 2010(Stevens et al., , 2016. However, less is known about the combined impact of O 3 and N, and contrasting results have been found. For grassland species and communities the interaction of O 3 and N has resulted in: no effect, damage limitation, and the increase of injury (Payne et al., 2011;Wyness et al., 2011;Bassin et al., 2013;Hayes et al., 2019). Responses are likely to vary depending on the species that make up the community and their sensitivity to O 3 . For example, Hayes et al. (2019) found that dune grasslands with high levels of N deposition (over a number of sites along an N deposition gradient) had the least damage caused by elevated O 3 , but that the plant community contained a high proportion of O 3 resistant forbs and grasses.
Leontodon hispidus (Rough Hawkbit) and Succisa pratensis (Devil's-bit Scabious) are important summer flowering temperate grassland species for pollinators in particular (Hicks et al., 2016;Ceulemans et al., 2017;Cole et al., 2017). For example, Succisa pratensis is the host plant for the threatened marsh fritillary (Euphydryas aurinia) butterfly eggs and larvae (Johansson et al., 2019). Adapted to low nutrient conditions, Succisa pratensis, faces threats from land use change, eutrophication and acidification (from the addition of fertilizers and atmospheric deposition of N and sulphur), and is therefore the subject of conservation efforts aiming to restore or maintain suitable habitat conditions for the plant (Vergeer et al., 2003a;van der Meer et al., 2014;Brunbjerg et al., 2017;Johansson et al., 2019). Although defined as a stress tolerator by Grime et al. (1988), elevated O 3 has been found to induce premature senescence in this species (Franzaring et al., 2000). Succisa pratensis vitality can also be damaged by eutrophication (from N), leading to problems for long term population viability (Pauli et al., 2002;Vergeer et al., 2003b), but the combined effects of increased O 3 and N are not known.
Leontodon hispidus has previously been found to be moderately sensitive to O 3 in terms of plant biomass at O 3 concentrations of over 40 ppb (Hayes et al., 2007(Hayes et al., , 2011Mills et al., 2009;Wilkinson and Davies 2009). In previous pot-based experiments Leontodon hispidus has displayed delayed senescence, reductions in root biomass (Hayes et al., 2011), and disruptions to stomatal control (Mills et al., 2009;Wilkinson and Davies 2009).
In this study we test growth and physiological responses (leaf cover, leaf litter, flowering, chlorophyll index, photosynthesis and stomatal conductance) of Leontodon hispidus and Succisa pratensis to ambient and elevated O 3 levels, with and without the addition of N. Irrigation occurred naturally from precipitation, and no supplementary watering was required apart from that associated with the N addition.

Ozone and nitrogen treatments
During the first season (May to September 2016) the plants were subjected to three O 3 treatments (Low, Medium and High) targeted to replicate the range of concentrations found in temperate zones, each with and without the addition of N (applied weekly in 5 L of water, to the equivalent of 40 kg N ha À1 yr À1 ). The level of N addition was chosen to be at the top end of the range of European N deposition (Stevens et al., 2010). Five litres of water were also applied to the zero N addition plots at the same time. Each O 3 treatment was replicated using three FAOE rings of 4 m diameter with 10 m spacing between the ring centres (Fig. 1).
The control treatment was randomly assigned within each row, and was upwind (of the prevailing wind direction) to the highest O 3 treatments. O 3 was supplied at a height of 30 cm with delivery achieved using an O 3 generator (G11, Pacific Ozone Inc. California, USA) supplied with oxygen concentrated from ambient air (Integra 10, SeQual Technologies Co. Ltd, Taiwan) and distributed to the rings via computer-controlled (LabView version 2012) solenoid valves (using pulse width modulation). Fans (200 mm, Xpelair, Southampton, UK) drove the O 3 through the delivery pipe (65 mm diameter, with 3 mm holes every 20 cm) at a rate of 0.17 m 3 s À1 . Wind speed (monitored using a WindSonic sensor, Gill Instruments Ltd, Lymington, UK) was used to instantaneously adjust solenoid operation, reducing O 3 release at wind speeds of less than 16 m s À1 and stopping delivery at speeds below 2 m s À1 . The O 3 concentration within each ring (at a height of 30 cm) was sampled for~3.5 min in every half-hour using an O 3 analyser (Thermo 49i, Thermo Fisher Scientific Inc. Massachusetts, USA). Air temperature (Skye Instruments Ltd, Llandrindod Wells, UK) on site was recorded with hourly averages calculated from 5 min interval raw data.

Leaf cover and leaf litter
Leaf cover of both plant species was determined in October 2016 using grids (containing 100 squares) overlain onto photographs of the plant blocks, with total leaf area and the proportion of damaged leaves assessed. The occurrence of frost damage was surveyed in each plot in January 2017 using the same method. Leaf litter was collected monthly from July 2016 to February 2017 (excluding December 2016) and dry weights (after oven drying at 60 C) were recorded.

Chlorophyll, photosynthesis and stomatal conductance
The following measurements were made on healthy leaves. The chlorophyll index (C.I.) was recorded weekly on both Succisa pratensis and Leontodon hispidus between 23 June and November 24, 2016 using a CCM200 m (Opti-Sciences, Hudson, USA). In addition, light-saturated photosynthesis (Asat) and stomatal conductance (gs) measurements were also recorded on Succisa pratensis and were made every two weeks (between 13 July and October 5, 2016) using a LI-COR 6400XT (Nebraska, USA) portable gas analyser. Leaves completely filled the chamber, and mean chamber conditions were: air temperature 20 C; incoming air CO 2 (carbon dioxide) concentration 400 mmol mol À1 ; light 1500 mmol m À2 s À1 (6400 LED source). Outgoing air relative humidity was within the range 60e80% and leaf vapour pressure deficit between 0.8 and 1.2 kPa.

Flowering
A count of the number of Leontodon hispidus flowering stems was made in August 2016, and Succisa pratensis in September 2018 (Succisa pratensis did not flower in 2016).

Statistical analysis
All ± margins reported show the standard error of the mean. Tests for statistical significance were carried out with R (version 3.6.3 R Core Team, 2020). Generalized linear models ('glmer', package 'lme4', Bates et al., 2015) with poisson distribution were used for count data (i.e. cover determined from number of squares, and number of flower stems). Linear mixed models (package'nlme', Pinheiro et al., 2017) were used for litter, C.I., Asat and gs data. Beta regression (package'reghelper', Hughes 2020) was used for proportional data (percentage leaf damage). Leaf cover, leaf litter and chlorophyll data were tested against the fixed factors of date, O 3 treatment, N treatment and species (as well as their interactions) using the random effect of FAOE ring and maximum likelihood (ML) estimation. Response variables relating to Asat, gs, and flowering (where data was recorded on Succisa pratensis alone or in separate years for each species) the fixed factors of date, O 3 treatment and N treatment (and their interactions) were used. Model residuals and plots were checked for the appropriateness of each model (including testing for overdispersion with 'glmer'), and data were transformed where necessary (cube root with litter data and square root with purple coloured leaf damage data). Where data contained repeated measures, date was included either as a factor (litter, Asat and gs) or as a continuous variable (days after start of measurements (DAS) for chlorophyll content) depending on model fit. The suitability of the inclusion of an autoregressive (AR) correlation term was determined by plots and model fit. The best model fit was assessed using comparisons of Akaike's Information Criteria (AIC). With the exception of beta regression, model results were summarised using ANOVA 'type 3' (package 'car', Fox and Weisberg 2011) which performs a Wald chi-square test (c 2 ). Tukey HSD post-hoc tests (package 'multcomp', Hothorn et al., 2008) and contrasts of means (package 'emmeans', Lenth, 2020) were carried out where appropriate.

Ozone treatments and climatic conditions
The seasonal mean O 3 levels over the exposure periods were~24 ppb for the Low treatment,~40 ppb for the Medium treatment and~57 ppb for the High treatment (Table 1). During the first season daily (24 h) mean levels reached 45 ppb, 66 ppb and 100 ppb in the Low, Medium and High treatments, respectively (Fig. 2a). Due to technical issues no O 3 was delivered to the Medium and High treatments between 7 and 9 August and 23e27 September 2016.
Climate conditions during the trial were within those generally expected for the site location in the western part of the UK ( Fig. 2b and c). The maximum air temperature on-site was 32 C recorded on July 19, 2016 (afternoon) and the lowest was 6 C on May 24, 2016 (early morning).

Plant response to treatments
3.2.1. Leaf cover and damage: significant effects of ozone, but not nitrogen Mean leaf cover (determined from count of number of squares) ranged from 3 (±6, High O 3 , added N) to 12 (±6, Low O 3 , no added N) for Leontodon hispidus, and from 61 (±6, High O 3 , no added N) to 74 (±4 Medium O 3 , no added N) for Succisa pratensis. No significant effect of N was found, however there was a significant interaction between O 3 and species (c 2 (2) ¼ 14.60, p < 0.001). Leontodon hispidus showed a trend of reducing leaf cover with increasing O 3 (p < 0.001, between Low and High O 3 ), whereas Succisa pratensis cover was slightly increased in the Medium O 3 treatment and slightly reduced in the High O 3 treatment this was not significant (Fig. 3). No leaf damage was noted for Leontodon hispidus. However, for Succisa pratensis the cover of damaged leaves (purple coloured, as a proportion of total cover) in the autumn increased with O 3 level (b ¼ À0.93, t (6) ¼ À4.47, p ¼ 0.004, Fig. 4a), although there was no significant effect of N or its interaction with O 3 . Whilst greater O 3 concentrations increased the proportional cover of frost damaged leaves for Succisa pratensis (b ¼ À0.70, t (6) ¼ À2.84, p ¼ 0.03, Fig. 4b), the addition of N did not significantly affect the level of damage (compared to those without added N).

Leaf litter: significant ozone and nitrogen effects
No significant effect of O 3 level on litter production was recorded for Leontodon hispidus. However, Succisa pratensis autumn litter mass was greater with increased O 3 , and at the highest O 3 treatment level the timing of the peak litter production was also affected (c2 (12) ¼ 37.63, p < 0.001, for month Â O 3 *species, Fig. 5). The increase in litter was only related significantly to plant cover at the Low O 3 level ( Figure SI 2).
Higher litter production occurred earlier with Leontodon hispidus than with Succisa pratensis. Peak litter production was in August for Leontodon hispidus where the mean weight of litter was 7.0 g (±1.0), 7.1 g (±1.5) and 7.2 g (±1.9) for the High, Medium and Low O 3 treatments (respectively). The peak mean weight of litter for Succisa pratensis was 8.4 g (±1.1) in November for the High O 3 treatment and 9.8 g (±1.2) and 5.5 g (±1.7) in January for the Medium and Low O 3 treatments (respectively).
Overall litter production was increased with added N (c 2 (1) ¼ 4.36, p ¼ 0.037), but there were no interactions with the other model terms. The total weight of litter over the measurement period was increased by 6% by the addition of N for Leontodon hispidus and by 31% for Succisa pratensis.

Chlorophyll index (C.I.): significant ozone and nitrogen interactions
The chlorophyll content varied with species and date. Leontodon hispidus C.I. remained between 30 and 40 for the measurement period, however C.I. for Succisa pratensis steadily increased from 30 to 50 from July to November (Fig. 6). C.I. for both species declined after a November peak. The mean maximum index values in November with and without the addition of N (respectively) were: 37 (±0.9) and 36 (±0.9) in the Low treatment, 36 (±1.0) and 37 (±1.3) in the Medium treatment, and 39 (±0.9) and 36 (±1.0) in the High treatment for Leontodon hispidus; and 48 (±1.1) and 46 (±1.0) in the Low treatment, 51 (±0.9) and 47 (±0.9) in the Medium treatment, and 49 (±1.2) and 48 (±0.8) in the High treatment for Succisa pratensis.
A significant interaction was found between date, species, O 3 , and N (c 2 (2) ¼ 9.38, p < 0.01). Leontodon hispidus C.I. was reduced in the Medium and High O 3 treatments compared to the Low treatment during October to November, with the addition of N resulting in an increase in C.I. in the highest O 3 treatment compared to the Medium and Low levels (Fig. 6a). Succisa pratensis C.I. was increased with increasing O 3 concentrations during August and September, and this was further increased by the addition of N (Fig. 6b).

Photosynthesis (Asat) and stomatal conductance (gs): significant ozone and nitrogen effects (Asat); significant ozone and nitrogen interaction (gs)
Light-saturated photosynthesis (Asat) and stomatal conductance (gs) were measured on Succisa pratensis only. Asat declined in the autumn compared to summer rates. The decline with time was enhanced at the High O 3 level (c 2 (10) ¼ 25.79, p ¼ 0.004) both with and without additional N (Fig. 7a). In the added N treatment Asat also reduced at a faster rate with time (c 2 (5) ¼ 12.77, p ¼ 0.03).
Gs followed the same trend as Asat by declining in the autumn compared to values during the summer. This decline was accelerated in the High O 3 treatment when there was no additional N. When N was added, gs was maintained across the different O 3 treatments (with a slight increase at High O 3 in October), although the seasonal decline was still present (c 2 (10) ¼ 21.38, p ¼ 0.02 (for date Â O 3 *N), Fig. 7b). Fig. 3. Leaf cover for both species derived from a count of the number of squares covered by leaves for each ozone treatment (Low, Medium and High), with and without the addition of nitrogen. Bars show the standard error. A significant interaction was found between O 3 and species (p < 0.001).

Fig. 4. O 3 and N impacts on a)
Proportional cover of damaged leaves (derived from the total count of number of squares covered by leaves) as indicated by their purple colouring (recorded in October 2016). The significance level is p < 0.001 obtained using square root transformed data; b) Cover of frost damaged leaves as a proportion of the total leaf cover, derived from the count of number of squares covered by the frost affected leaves (recorded in January 2017). The significance level is p < 0.05. Treatments shown are ozone concentrations (Low, Medium, and High), with and without the addition of nitrogen. Different letters denote significance between the three ozone treatments and bars are standard error.
A. Holder, F. Hayes, K. Sharps et al. Global Ecology and Conservation 24 (2020) e01345 3.2.5. Flowering: significant effect of ozone, but not nitrogen The number of Leontodon hispidus flowering stems (recorded in 2016) was similar across the treatments with no effect of O 3 or N treatments (Fig. 8a). Succisa pratensis did not flower in 2016 and the number of flowering stems two years later were  A. Holder, F. Hayes, K. Sharps et al. Global Ecology and Conservation 24 (2020) e01345 reduced in the elevated O 3 treatments (c 2 (2) ¼ 13.41, p ¼ 0.001, Fig. 8b). There was no significant effect of the different N levels, or the interaction between O 3 and N treatments.

Discussion
This study has shown that the important grassland speciesare negatively affected by elevated O 3 even at a moderate level (seasonal mean 40 ppb), with few impacts exacerbated by elevated N.
Small decreases in leaf cover and chlorophyll index were observed with increased ozone for Leontodon hispidus, with the addition of N resulting in an increase in chlorophyll index only at the highest O 3 level and resulting in an overall increase in litter production. However, no effect of O 3 or N on flowering were found during the first season. Other studies into the effects of elevated O 3 with Leontodon hispidus have reported delayed senescence, root biomass reductions (Hayes et al., 2011), and alterations in stomatal control (Mills et al., 2009;Wilkinson and Davies 2009). It is, therefore, possible that the species was affected in aspects not measured (e.g. photosynthesis or below ground biomass).
A stronger effect of both O 3 and N treatments was seen with Succisa pratensis. After only three years of exposure to O 3 pollution, at concentrations similar to recent levels recorded across some European temperate grassland areas (European Fig. 7. Treatments shown are ozone concentrations (Low, Medium, and High), with and without the addition of nitrogen. Bars are standard error. a) Succisa pratensis mean light-saturated photosynthesis (Asat) for each treatment. The interactions between date and O 3 , and between date and N are significant at p < 0.05; b) Succisa pratensis mean stomatal conductance (gs) for each treatment. The interaction between date, O 3 and N is significant at p < 0.05. Environment Agency (EEA), 2017), flowering of Succisa pratensis was reduced suggesting O 3 is already causing a reduction in pollinator resources (although we acknowledge that there may have been other interacting factors not taken into account, e.g. climate or pest species). Even though species like Succisa pratensis, with low specific leaf areas and relative growth rates, have been identified as more stress tolerant (Grime et al., 1988), our findings agree with the results of an earlier study into the effects of O 3 on the plant. In a pot based, open top chamber experiment Franzaring et al. (2000) observed that elevated O 3 (seasonal maximum of 77 ppb) did not influence growth (biomass) but that premature senescence was recorded with leaf litter significantly affected. In this current trial, whilst leaf cover was not impacted to a great extent, there was a strong effect of O 3 on leaf litter production and leaf damage caused by frost after only one season's exposure.
The increasing chlorophyll content with elevated O 3 implies that more resources were allocated to the newest leaves as the older ones died back prematurely and fell as litter (measurements were made on replacement healthy leaves). During the summer months resources were prioritised to new leaves so that stomatal conductance and photosynthesis could be maintained, but during the autumn this was not sustained and stomatal conductance declined with increasing O 3 concentrations. However, the decline in stomatal conductance observed at high O 3 was mitigated by adding N. Slow or impaired stomatal response due to elevated O 3 has been found at times for other plants including grassland species (Mills et al., 2009. The rate of photosynthesis is normally higher during the summer/flowering months and this was true for all of the treatments, but it should be noted that this also corresponds with the period when ambient O 3 levels are typically higher. Succisa pratensis is therefore susceptible to uptake of O 3 through stomatal fluxes during high ozone episodes. Although senescence occurred in October for all treatments, we noted a steeper decline in the rate of photosynthesis in the higher O 3 treatments during September and October as winter die-back was accelerated. This occurred both with and without additional N. These stress responses are likely to cause a decline, or change in allocation of, assimilated photosynthates with detrimental effects for plant health in subsequent years (Cooley and Manning 1987;Mills et al., 2013). This could in turn have implications for species that depend on the plant's health, density, and food quality (i.e. nectar, pollen and green leaves) (Tjørnløv et al., 2015;Brunbjerg et al., 2017;Ceulemans et al., 2017;Ghidotti et al., 2018). Detrimental impacts on plant health and ability to cope with environmental stresses were evident in the increased leaf damage and reduced flowering for plants that were exposed to elevated O 3 . This has further consequences for the long-term viability of populations where reducing numbers of Succisa pratensis can reduce genetic diversity (Vergeer et al., 2003a). Furthermore, elevated O 3 increased the weight of litter in the Medium and High treatments (regardless of the amount of plant cover) from mid-August onwards. A lower number of green leaves during late summer will impact food resources of the larvae of the marsh fritillary (Euphydryas aurinia) butterfly that, almost exclusively, use the plant as host for food and form silken protective and hibernation webs amongst the leaf litter (Meister et al., 2015: Tjørnløv et al., 2015Johansson et al., 2019). The larvae have limited mobility and whilst they are able to move to neighbouring plants this exposes them to predation and adverse weather conditions. Anthes et al. (2003) noted that the few larval groups (of Euphydryas aurinia) observed to move to another plant were those that had completely defoliated their host plant.
In contrast to Leontodon hispidus, the weight of Succisa pratensis litter remained higher in the elevated O 3 treatments during autumn and into winter, even though O 3 treatments stopped at the end of September. This prolonged reduction in plant health and early senescence may also influence the plants ability to compete with other grassland species, thereby reducing habitat quality and biodiversity (Vergeer et al., 2003a). This change in microhabitat structure may result in less suitable locations for egg placement by Euphydryas aurinia females who prefer sites with less dense vegetation that allows light to penetrate to lower layers (Pielech et al., 2017). The balance of leaf cover between the two species tested was not influenced by elevated O 3 or added N suggesting that (at the levels added) these treatments did not affect direct competition between them.
Increased litter also has the potential to impact on seed germination and establishment, as well as the microhabitats. Hovstad and Ohlson (2008) found that the addition of 900 g litter m À2 produced a negative response in Succisa pratensis seedling emergence, and the amount of litter from competitor species also negatively affected Succisa pratensis in another grassland study (Hurst and John 1999). Extra carbon (C) and N released from greater amounts of litter may favour competition from other plant species and, in the potential change to soil C and N ratios, temperature, and moisture status, alter microbial activity (Steinbeiss et al., 2008;Kuzyakov 2010;Lange et al., 2015).
To our knowledge this is the first published study including the interaction of O 3 and N for a forb grassland species adapted to low nutrient conditions. Eutrophication has previously been shown to damage viable populations of Succisa pratensis by either reducing biomass (Pauli et al., 2002), or conversely by increasing biomass and seed production but reducing reproductive fitness and germination success (Vergeer et al., 2003b). In this study, the addition of N had no visible effect on either Leontodon hispidus or Succisa pratensis plant cover or flowering (although Succisa pratensis flowering was recorded two years after the N treatment was applied when it is less likely to have had an impact). However, N addition accelerated the developmental cycle of Succisa pratensis and produced a higher litter quantity for both species. The N treatment applied (40 kg N ha À1 yr À1 ) was twice the amount suggested as the point at which N deposition has been shown to reduce growth (20 kg N ha À1 yr À1 , Payne et al., 2013), but was within the European range of between 2 and 44 kg N ha À1 yr À1 (Stevens et al., 2010).
Climate change-induced earlier vegetation growth in spring is exacerbated by N deposition, and can cause a cooling of the microclimate in the immediate zone of spring developing larvae with negative implications for butterfly populations (Wallis de Vries and Van Swaay, 2006). It is not known how the increase of litter into the immediate area of the hibernating larval web would interact with this cooling issue, but alterations in humidity and temperature are likely.
Whilst some negative impacts have been documented as part of this study there may be other detrimental effects that will have implications for the grassland ecosystem. For example, in a study of grass and buttercup species the interaction of O 3 and N caused negative effects on root biomass , a possible outcome not measured as part of this trial. O 3 pollution has also been shown to disrupt plant to plant, plant to herbivore, and plant to pollinator signalling for a number of species through reducing the lifetime of biogenic Volatile Organic Compounds (BVOCs) or by causing the emission of stressinduced BVOCs (Llusi a et al., 2002;Pinto et al., 2010;Owen and Peñuelas, 2013;Farr e-Armengol et al., 2013). It is known that caterpillars feeding on Succisa pratensis induce emissions of herbivory BVOCs that influence the behaviour of nearby plants as well as herbivores and their predators (Peñuelas et al., 2005). Indications are that herbivory BVOCs are degraded by O 3 pollution (Pinto et al., 2007;Fuentes et al., 2013), but it is not known if O 3 affects Leontodon hispidus or Succisa pratensis signalling in this way or whether other floral signalling mechanisms are affected. O 3 air pollution continues to be a problem for natural vegetation and crops (Yan et al., 2019;Ronan et al., 2020) and this study showing the impact on a high value plant for pollinators further highlights the need for concerted action at the global scale to reduce pre-cursor O 3 emissions alongside habitat management efforts.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.