†Author to whom correspondence should be addressed. E-mail: email@example.com
1Atlantic bryophytes are of European conservation importance, yet the effect on them of excess atmospheric nitrogen is relatively unknown. This study assesses the effects of increased atmospheric N deposition on the growth and tissue N of epiphytic Atlantic bryophytes, and their potential to recover following a decline in N deposition.
2The N received in stemflow by bryophytes at two sites was measured and compared to model predictions.
3Four species of epiphytic bryophytes (Isothecium myosuroides, Dicranum scoparium, Frullania tamarisci and Ulota crispa), typical of Atlantic Oak woods, were studied in a 12-month reciprocal transplant experiment between a pristine Oak woodland receiving a modelled atmospheric deposition of 12 kg N ha−1 year−1 and a polluted one receiving 54 kg N ha−1 year−1.
4Tissue N concentration increased and growth declined following an increase in atmospheric N deposition in all species except Ulota crispa. Conversely, tissue N concentration decreased and growth increased in Frullania tamarisci following a decrease in atmospheric N deposition, with similar non-significant patterns in the other species.
5The reciprocal transplants indicate a detrimental effect of increased N deposition on the bryophyte species studied. The study indicated recovery following a decrease in atmospheric N deposition, but the responses caused by decreased N deposition were smaller than those due to increased N deposition. This suggests that the time-scale for recovery of bryophytes from excess N deposition is longer than the timescale of nitrogen impacts.
Epiphytic lichens are well known as indicators of air pollution (Hawksworth & Rose 1970), with changes in the community composition of lichens used to assess exposure to SO2, NH3 and NOx (van Dobben & De Bakker 1996; van Dobben et al. 2001). Bryophytes have been used less commonly as indicators of air pollution, although there is ample evidence that they can be used for this purpose (Rao 1982). Bryophytes appear to be particularly sensitive to increased atmospheric N deposition (Woodin, Press & Lee 1985; Pitcairn, Fowler & Grace 1995). Potential responses of bryophytes to increases in air pollution may be seen in three ways: changes in community composition; changes in growth rate; and changes in concentration of the pollutant in bryophyte tissue. The latter two changes may provide earlier indicators than changes in community composition.
Increased atmospheric N is detrimental to certain bryophyte species. New shoot growth in Polytrichum formosum was reduced when exposed to 122 µg NO2 m−3 for 37 weeks, a N concentration similar to those found on rural roadside verges (Bell, Ashenden & Rafarel 1992). The growth of Sphagnum cuspidatum was reduced by the addition of 0·01 mm or to the water in which it was growing (Press, Woodin & Lee 1986). Growth of Hylocomium splendens and Pleurozium schreberi was strongly reduced by the addition of 30–60 kg N ha−1 year−1 (Dirkse & Martakis 1992), and Racomitrium lanuginosum growth was reduced by 58% with an additional 10 kg N ha−1 year−1 on mountain plots in north-east Scotland (Pearce & van der Wal 2002).
Previous work on N deposition and bryophytes has concentrated on terricolous bryophytes. By contrast, this study considers epiphytic bryophytes characteristic of Atlantic Oak woods which occur in extremely oceanic, cool temperate conditions characterized by very high precipitation throughout the year. The Atlantic bryophyte flora is of particular conservation importance (UK Biodiversity Group 1995; Hodgetts 1997), and plays an important role in the ecology of these woods (Rieley, Richards & Bebbington 1979), yet the sensitivity of this group to N pollution is unknown. While most other studies in the UK have concentrated on more polluted sites, this study assesses the influence of atmospheric N deposition on growth and tissue N concentrations of epiphytic bryophytes in relatively unpolluted and polluted Atlantic Oak woods. A replicated, manipulative experimental study was designed to test the following hypotheses:
1Growth of epiphytic bryophytes declines when atmospheric N deposition increases;
2Tissue N concentration of epiphytic bryophytes increases when atmospheric N deposition increases.
The recovery of epiphytic bryophytes following a decline in the concentrations of deposited N pollutants has not previously been assessed. However, if current and future efforts to reduce N pollution concentrations are to be beneficial, we need to know if, and how, the epiphytic communities will respond. The study therefore aimed to test the following additional hypotheses:
3Growth of epiphytic bryophytes increases when excess atmospheric N deposition decreases;
4Tissue N of epiphytic bryophytes decreases when atmospheric N deposition decreases.
Materials and methods
Two Atlantic Oak woodlands were chosen: Ariundle, a pristine site in western Scotland near Strontian, Fort William (latitude 56°44′ N, longitude 5°32′ W), and Borrowdale, a polluted site, in the Lake District, England (latitude 54°30′ N, longitude 3°32′ W). Both sites are ancient Oak woods previously studied for their epiphytic communities (Farmer, Bates & Bell 1991; Bates 1992a). Nitrogen deposition rates for 1995–97 were estimated at 12 and 54 kg N ha−1 year−1 for Ariundle and Borrowdale, respectively (Smith et al. 2000; NEGTAP 2001).
Four epiphytic bryophyte species were studied: Isothecium myosuroides, Dicranum scoparium, Ulota crispa and Frullania tamarisci (nomenclature follows Paton 1999 for liverworts and Smith 1978 for mosses). Isothecium myosuroides is a trunk-dwelling pleurocarpous moss with a subdendroid growth form. Dicranum scoparium is an acrocarpous moss, and F. tamarisci is a foliose liverwort which forms prostrate mats; both species are epiphytic and terricolous, but only epiphytic samples occurring on the trunks were used in this experiment. Ulota crispa is a small, acrocarpous moss (5–10 mm tall) which grows as small cushions on twigs. These species were selected as they were sufficiently common for samples to be taken without a detrimental effect on the populations at the sites, and their growth form enabled clumps to be taken and transplanted.
A replicated transplant experiment was set up between the two sites with the four species. There were four types of transplant:
NT, not transplanted: no change in N supply and not transplanted;
T0, transplant, no change in N supply: the bryophyte was removed from the tree, tagged and reattached to the same place on the same tree;
TI, transplant, increased N: bryophyte was transplanted from Ariundle (low N) to Borrowdale (high N);
TD, transplanted, decreased N: bryophyte transplanted from Borrowdale (high N) to Ariundle (low N).
Each site acted as both a source site (where the bryophyte came from) and a host site (where the bryophyte was transplanted to). This design allowed the effect of moving the bryophyte (transplant type) to be assessed independently of the effects of source and host site. The entire experiment was conducted on Oak trees (Quercus robur L., Q. petraea (Matt.) Liebl. and the presumed hybrid Quercus × rosacea Bechst.)
The trunk-dwelling epiphytes (I. myosuroides, D. scoparium and F. tamarisci) were reattached to trees using small nets made of double-threaded, 26 µm diameter nylon thread with a mesh size of 9 mm when stretched out. The bryophytes were placed inside the nets, which were stapled to the trunk of the tree. For the T0 samples the bryophytes were reattached in exactly the same place; for the TD and TI samples the bryophytes were reattached on trees at the same aspect and height as at the source site. Where possible, TD and TI samples were transplanted into the middle of a clump of the same species already on the tree, to minimize effects of neighbour removal (the samples having been taken from the middle of larger clumps on the source sites). For D. scoparium there was insufficient material at Borrowdale to provide transplant material, so only NT samples were established. Unfortunately these were destroyed, probably by sheep, during the course of the experiment. Other NT samples of D. scoparium were collected from Borrowdale for tissue N analysis, but as these had not been tagged they could not be included in the growth analysis. Transplants were left for 1 year: the experiment was set up in February 2001 and samples collected at the end of January 2002.
Ulota crispa is a twig-growing epiphyte, and therefore could not be transplanted in the same way as the other species. T0, TD and TI transplant types were established; twigs with U. crispa growing on them were removed from the tree, the moss cushions measured and the twigs reattached either onto the same tree (T0), or onto a tree at the other host site (TD and TI). Twigs were reattached by tying them onto other twigs with string. As it was impossible to measure clump size accurately while the twigs were still attached to the tree, and because this technique (in contrast to that used for the other species) did not remove the bryophyte from its substrate, NT transplants were not carried out. Ulota crispa grows on dead twigs, therefore the removal of twigs from the tree should not have affected growth.
For each of the three trunk-dwelling epiphytes, eight replicate bryophyte clumps were measured for each transplant type. On each clump, eight shoots were tagged 10 mm from the apex of the shoot with a single thread of embroidery cotton (Jonsdottir, Crittenden & Jagerbrand 1997). After a year the samples were collected and the shoots measured to an accuracy of 0·25 mm using callipers. Once the tagged shoots had been measured they were removed from the clump, air dried at 30 °C, and weighed. This allowed the mass per mm shoot to be calculated. The length and width of 21–26 replicate U. crispa clumps for each transplant type were measured at the start and end of the experiment to provide a measure of growth.
tissue n measurements
At the start of the experiment, five random samples of each of the four bryophyte species from each site were collected. These were analysed for tissue N to provide baseline data. At the end of the experiment all bryophyte samples in the transplant experiment were analysed for tissue N.
After collection samples were sorted to obtain a pure sample of the desired species, and washed with deionized water to remove any dry N deposition on the surface of the samples (Leith et al. 2001). The samples were then oven-dried at 70 °C and hammer-milled to <0·8 mm. The ground powder was analysed for total N by combustion (CNS Analyser, Elementar Model: Vario EL; Burkard Scientific Ltd, Uxbridge, UK). Due to the small size of the samples once they had been washed, sorted and ground, many replicates within transplant types had to be combined, resulting in a sample size of fewer than eight and, in one case, one large pooled sample (see Fig. 2 for details).
At each site, stemflow (rainwater down the tree trunks) was measured on four trees not used for the transplant experiment. This allowed a direct estimate to be made of the N received by the epiphytes. The stemflow was collected at a height of 1·5 m above ground and stored in 99 l black polythene bins. Thymol was added to inhibit transformation of N ions in the water. The volume of stemflow water was measured monthly, and samples were collected and analysed for and (I.D. Leith et al., unpublished, give details of methodology). The stemflow-measuring equipment was set up at the start of the transplant experiment (January 2001). However, sample collection had to be abandoned from February until August 2001 due to restrictions imposed by the 2001 outbreak of foot-and-mouth disease in the UK. This means that stemflow data are not available for the entire period of the transplant experiment.
Growth data were analysed by fitting linear mixed models to individual shoot-level data (MIXED procedure in SAS ver. 8·1) (SAS 1999). The growth data were normalized using a square-root transformation. Tree and clump were included as random effects, and the height and aspect (as northerly and easterly components) of the clump on the tree were included as covariates. Species, source site, host site and transplant type (NT, T0, TI, TD) were included as factors. This method, using individual shoot data, is a more efficient statistical method than using mean growth per clump. It allowed us to handle the different number of shoots per clump (some shoots were lost); issues of pseudo-replication were taken care of by including clump as a random effect.
Weight per shoot length was also analysed by fitting linear mixed models with tree included as a random effect and species, source site, host site and transplant type included as fixed effects. Ulota crispa growth was analysed in the same way as the other growth data, but with twig included as a random effect, as there was often more than one U. crispa clump per twig. Source site, host site and transplant type (T0, TI, TD) were included as factors.
Tissue N for all four species was analysed by fitting generalized linear mixed models to individual clump data (Proc GLM procedure in SAS; SAS 1999). It was not possible to include the random effects of tree and clump, as samples had to be pooled within transplant types in order to obtain enough material for chemical analysis. For both tissue N and growth, three analyses were carried out: (1) to test if there was an effect of transplantation, data from NT and T0 transplants were analysed; (2) data from both source sites were included in the analysis; (3) data from each source site were analysed separately. Differences in stemflow chemistry between the two sites were also analysed by fitting generalized linear mixed models.
Over the year of the experiment, I. myosuroides, D. scoparium and F. tamarisci shoots grew between 1 and 4 mm, depending on species and transplant type (Fig. 1). Tree, and height and aspect of the clump, had no effect on growth and were omitted from the final analysis. However, clump was included as a random variable within the analysis, as there was variation in the growth of shoots depending on the bryophyte clump in which it grew. For bryophytes that had received no change in N input, there was no significant difference in growth between those that had been transplanted and those that had not (comparison of NT and T0 transplant types). Analysis of data from both source sites combined showed that host site (F1,590 = 11·03, P = 0·05) and species (F2,590 = 11·44, P < 0·0001) had a significant effect on growth, with those bryophytes hosted at Borrowdale growing significantly less than those at Ariundle. Transplant type (NT, T0, TD, TI) and source site had no significant effect on growth. The data were then analysed separately by source site. Bryophytes sourced from Ariundle grew less when hosted at Borrowdale (TI), than at Ariundle (T0 or NT) (Fig. 1). Species (F2,369 = 4·42, P = 0·05) and host site (F1,369 = 13·18, P = 0·001) had a significant effect on growth, but transplantation type did not. Only species had a significant effect on the growth of samples sourced from Borrowdale (F1,225 = 12·21, P = 0·001). However, F. tamarisci and I. myosuroides sourced from Borrowdale showed different growth patterns: I. myosuroides had similar growth (2–2·5 mm) irrespective of host site, while F. tamarisci grew significantly more when hosted at Ariundle (TD) than when hosted at Borrowdale (NT or T0) (F1,102 = 5·1, P = 0·05). Transplant type, host site and source site had no significant effect on weight per mm shoot of F. tamarisci, I. myosuroides and D. scoparium, or on the size (length, width) of U. crispa clumps.
At the start of the experiment there was a significant difference between sites in the tissue N concentration of the four bryophyte species (F1,24 = 63·21, P < 0·0001). Tissue N was between 0·3 and 1% greater in samples from Borrowdale than those from Ariundle (Fig. 2). There was no significant yearly variation in tissue N between those samples taken at the start of the experiment (February 2001) and those (T0 and NT) taken a year later in January 2002. Analysis of NT and T0 transplant types showed that transplantation had no significant effect on tissue N concentrations.
Analysis of all transplant types at both sites showed that there was a significant effect of species (F3,40 = 7·53, P = 0·001), source site (F1,40 = 237·31, P < 0·0001) and host site (F1,40 = 60·81, P < 0·0001) on tissue N concentrations. There were also significant interactions between source site and species (F3,40 = 13·4, P < 0·0001) and between host site and species (F3,40 = 11·47, P < 0·0001) The data were then analysed separately by source site. Species (F3,28 = 6·95, P = 0·01) and host site (F1,28 = 101·54, P < 0·0001) had a significant effect on tissue N for those samples sourced from Ariundle, and there was also a significant interaction between species and host site (F3,28 = 11·01, P = 0·0001). Dicranum scoparium, I. myosuroides and F. tamarisci samples had between 0·2 and 0·4% greater tissue N concentration if they were hosted at Borrowdale (transplant type TI), than if they were hosted at Ariundle (transplant type T0 or NT) (Fig. 2). Isothecium myosuroides, F. tamarisci and U. crispa sourced from Borrowdale had smaller tissue N concentrations if they were hosted at Ariundle (transplant type TD) than if they were hosted at Borrowdale (transplant types T0 and NT) (Fig. 2), but this was not significant. This was because the species varied in their response to a decline in atmospheric N, with tissue N declining significantly in F. tamarisci (F1,6 = 8·81, P = 0·05) but not in I. myosuroides. It was not possible to analyse the results from U. crispa separately as the samples had to be pooled in order to obtain enough material for analysis.
stemflow volume and chemistry
Stemflow data are presented as volume of water per area of tree trunk (l m−2) where the area is calculated based on the diameter of the tree at breast height (d.b.h.). Mean monthly stemflow at Ariundle ranged from 35 to 1155 l m−2, and at Borrowdale from 85 to 2021 l m−2 (Fig. 3a), but was not significantly different between sites. There were, however, significant differences between months (F5,36 = 3·0, P = < 0·05) and a significant site × month interaction (F5,36 = 4·14, P < 0·01). There was a significant difference between sites in both and stemflow concentrations (for , F1,34 = 21·12, P < 0·0001; for , F1,35 = 5·42, P < 0·05), but no significant effect of month or interaction between month and site. and concentrations in the stemflow were always significantly smaller at Ariundle (monthly means ranging from 1 to 3·8 µmol l−1, 0·3–6·8 µmol l−1) than at Borrowdale (monthly means 1·8–64 µmol l−1, 6·3–60 µmol l−1) (Fig. 3b,c). A total stemflow flux of 3 mmol per tree and 0·4 mmol per tree was recorded in the stemflow at Ariundle over the 6 months, with 10 mmol per tree and 5 mmol per tree recorded at Borrowdale.
growth and tissue n
Due to difficulties in directly manipulating the atmospheric N deposition at any one site, a reciprocal transplant experiment is a convenient way to assess the effect of changes in atmospheric N deposition on bryophyte growth. However, this introduces many potential confounding factors, such as the effect of transplantation and climatic differences between the two sites. The results of the present experiment showed that there was no effect of transplantation. Source site also had no effect on growth, but host site did, therefore it was the conditions at the host site during the year of the experiment that affected the growth of bryophytes.
The most relevant climatic differences between Ariundle and Borrowdale that could affect growth are precipitation and temperature. The average annual precipitation and maximum temperature at the two sites are similar, but Ariundle has slightly lower minimum temperatures (Table 1). One would expect growth to be greater at Borrowdale than Ariundle, as bryophytes grow better under wet, warm conditions (Pitkin 1973), yet bryophyte growth was greatest at Ariundle so the climatic differences between the two sites do not explain the differences in growth. Differences in irradiance is unlikely to be the cause of the differences in growth at the two sites as they had similar canopy cover: Ariundle 96·7% and Borrowdale 95·6% (measured using a hemispherical densiometer in early September 2001).
Table 1. Climatic data from meteorological stations near study sites (distance and direction from study site in parentheses)
Annual precipitation (mm)
Average annual maximum daily temperature (°C)
Average annual minimum daily temperature (°C)
Data from British Atmospheric Data Centre (BADC) Meteorological Office Records.
Rosthwaite (3 km NE)
Seatoller (1 km NW)
Drumnatorran (2 km SW)
Bellsgrove (1 km NE)
Onich (20 km E)
Inverailort (20 km NW)
Other factors that may influence bryophyte tissue N are bark chemistry (Brown 1982), season (tissue N being negatively correlated with moisture levels and rainfall), and bryophyte morphology (Bakken 1995b). However, these factors can be eliminated from this study. Analysis of bark chemistry at the two sites (data not presented) showed no significant difference in total P, N or pH. The samples from the different sites were collected at the same time of year, eliminating season. The weight per mm of shoot provides a measure of the structure of the moss: the greater the weight, the more side branches are present. As the morphology of the bryophytes did not change with host site, there is no evidence that apical shoots of bryophytes hosted at Borrowdale were growing less because the side shoots were growing more, or that differences in tissue N are due to differences in the ratio of main stem to side branches, which may have different tissue N levels to the main stem (Bakken 1995b).
atmospheric n deposition
Assuming that the 6 months’ stemflow measurements reflect differences for the entire year of the experiment, during the year the bryophytes hosted at Borrowdale received a stem flux of 14 mmol per tree and 9·2 mmol per tree−1 more than those at Ariundle. These site-based measurements confirm the previous model predictions of greater atmospheric deposition at Borrowdale compared with Ariundle (Smith et al. 2000; NEGTAP 2001). While the effect of the 2001 outbreak of foot-and-mouth disease will have been minimal for Ariundle, at Borrowdale foot-and-mouth is expected to have reduced the dry deposition of NH3 by as much as 30–50% compared to other years (Sutton et al. 2002).
The ratio of total N ( + ) measured in the stemflow (15/3·4 mmol per tree) was 4·4 (Borrowdale/Ariundle), while the ratio for modelled N deposition (Smith et al. 2000; NEGTAP 2001) (54/12 kg N ha−1 year−1) was 4·5. Due to canopy interactions of deposited and and different apportionment between stemflow and throughfall between the two sites one would not expect the ratios to be exactly proportional, but this demonstrates a close consistency between stemflow and modelled deposition. It should be noted, however, that NO3− accounts for a larger contribution (40%) in the modelled estimates (Smith et al. 2000; NEGTAP 2001).
effects of increased n
The differences in modelled N deposition and measured N in stemflow between Borrowdale and Ariundle indicate a clear difference between the sites that is the likely cause of the decline in growth. This is consistent with other studies of bryophytes in other habitats (Press et al. 1986; Bell et al. 1992; Dirkse & Martakis 1992; Potter et al. 1995; Pearce & van der Wal 2002). Possible indirect mechanisms by which increased N deposition affects growth include competition for light and water, but neither of these are issues at these sites due to the epiphytic nature of the plants. Direct effects of N on plants include direct toxicity and nutrient imbalance caused by cation exchange. Increased N has direct toxic effects on R. lanuginosum and causes K leakage from the cells (Pearce, Woodin & van der Wal 2003). Further work is necessary to confirm the mechanism(s) by which growth is inhibited.
While tissue N concentration increased when plants went from Ariundle to Borrowdale, the tissue N was still less than that in plants sourced and hosted at Borrowdale. Thus there is a time lag before bryophyte tissue N corresponds to the atmospheric N inputs (Bakken 1995b), explaining why both source and host site were significant in explaining tissue N. While bryophyte tissue N and growth clearly do respond to increases in atmospheric N, it would be of interest in future studies to measure the rate at which they respond to both increased and decreased atmospheric N deposition.
Unlike the other species, neither the growth nor the tissue N of U. crispa showed any effect of increasing N. As a twig epiphyte, U. crispa may receive a different amount of N than the trunk epiphytes, and may thus respond differently from the other species in this experiment. However, U. crispa is sensitive to pollution (Grubb, Flint & Gregory 1969) and the lack of increase in tissue N is surprising. The lack of significant differences in growth could be due to a lower sensitivity of U. crispa, as well as problems of measuring clump size accurately. A single year may not be sufficient time to detect changes in growth using this method.
recovery of bryophytes following a decline in atmospheric n
Bryophytes hosted at Ariundle had greater growth and lower tissue N than those hosted at Borrowdale. Frullania tamarisci showed the greatest signs of recovery, having significantly greater growth and smaller concentrations of tissue N when atmospheric N levels declined. As a liverwort, F. tamarisci has a thinner growth structure and hence a greater surface area to volume ratio, and may respond more rapidly than the mosses to changes in atmospheric N. Frullania tamarisci may also have a smaller source pool of N in terms of dead shoots and cells from which to recycle N, compared with I. myosuroides. Growth of I. myosuroides showed no increase with a decrease in atmospheric N, suggesting that it was already growing at its maximum rate at Borrowdale and could not respond to a decrease in atmospheric N. Tissue N of I. myosuroides declined, but not significantly so following a reduction in atmospheric N. Tissue N did not fall to concentrations comparable to those of species sourced and hosted at Ariundle. The slow rate of decline in tissue N following a decrease in atmospheric N may be because the bryophytes can recycle N, either within the shoot or within the bryophyte clump, from senescent or dead cells in other parts of the plant (Brown 1982; Bakken 1995a, 1995b), enabling them to maintain high tissue N concentrations for a considerable time following transplantation to a site receiving lower inputs of atmospheric N. Nonetheless, partial recovery in less than 1 year may be considered rather rapid. The tissue N of epiphytes may respond more quickly than that of vascular plants to changes in atmospheric N deposition, as the pool of N available to epiphytes is smaller than that available to plants in the soil. The recovery for epiphytes shown here is in contrast to the limited and very slow recovery from N deposition for terricolous plant communities (NEGTAP 2001).
Recovery of epiphytic bryophytes following a decline in atmospheric N has not previously been studied. These results show that recovery can occur following a decrease in atmospheric N deposition, in terms of increased growth and decreased tissue N for F. tamarisci, and trends of decreased tissue N for I. myosuroides and U. crispa. Although recovery of epiphytes following a decrease in N is possible, the responses in time due to decreased N are not as great, or as fast, as those due to increased N. To clarify these effects and time-scales further, a more detailed study over a longer time period is required.
This work was funded by a NERC small GANE grant, reference NER/T/2000/00037. We thank Scottish Natural Heritage and the National Trust for use of their sites; Maurice Parkhust of the National Trust for collecting the monthly water samples at Borrowdale; Maddie Thurlow for help with fieldwork; David Fowler and Rognvald Smith (CEH Edinburgh) for help and advice; and David Elston of BioSS for statistical support.