THE SEA EMPRESS INCIDENT AND THE LIMPETS OF FRENCHMAN S STEPS, TWENTY YEARS ON. JOHN ARCHER- THOMSON

Transcription

1 ARCHER- THOMSON (216). FIELD STUDIES ( studies- council.org/) THE SEA EMPRESS INCIDENT AND THE LIMPETS OF FRENCHMAN S STEPS, TWENTY YEARS ON. JOHN ARCHER- THOMSON Honey Hook Cottage, Lower Freystrop, Haverfordwest, Pembrokeshire, SA62 4ET. UK Students and staff from FSC Dale Fort Field Centre have studied limpet populations on Frenchman s Steps shore for over thirty years. Variations in population density and age structure have been measured and in particular the effects of the Sea Empress oil spill have been investigated. The oil pollution reduced the population density significantly and affected smaller (younger) limpets in particular, especially those on the lower half of the shore. Since then further changes in limpet population density have occurred. These are interpreted in the light of long- term oscillations in populations of barnacles and dogwhelks, which are related to the use (and subsequent banning) of Tri- Butyl- Tin anti- fouling paint. Long- term monitoring, using data collected by student groups, can be useful in highlighting variations in population density over time as well as being a worthwhile educational experience in its own right for the students and staff involved in the data collection. FIGURE 1. Frenchman s Steps, Pembrokeshire (Grid Ref. SM82253). INTRODUCTION Limpets of the Genus Patella are very important organisms on European rocky shores. Their differential grazing activity on a wide range of microorganisms (biofilm) and seaweeds, including Fucus spp., and encrusting red seaweeds, affects the shore community so significantly that limpets have been called a keystone species in that their effect on community composition is greater than would be predicted from their abundance or biomass alone (Jenkins et al., 1999, Little et al., 9). Factors that affect limpet abundance are therefore likely to affect the shore as a whole, which makes any long- term monitoring programme of limpets potentially interesting. Meaningful statements about the effect of an event on a rocky shore population depend on knowledge of what the population s status was before the event. It is also imperative to know about natural variations in the population before deductions can be made concerning unnatural ones. However, very few suitable long- term data sets exist, for any ecosystem, to allow such deductions to be made with any confidence. When the Sea Empress oil spill occurred (see Figure 2) in February, long- term data sets on the numbers and size- range of the common limpet Patella vulgata L, collected by student groups from FSC Dale Fort, did exist. Details of the methodology are given below. Comparisons were made between pre- and immediately post- pollution results and the conclusions were presented in Field Studies (Archer- Thomson, 1999). Useful background information on the effects of oil spills on rocky shores, the Sea Empress oil spill, chemical details of the oil and the biology of limpets can be found in this paper, but see Branch (1981) for a comprehensive overview of limpet biology and Crump et al., 1 Field Studies Council (2/9/216)

2 (1998) for an account of the oil spill on West Angle Bay, Pembrokeshire. In summary, the oil spill reduced the numbers of limpets in total (Figure 3) with especially noticeable declines in smaller size classes on the lower part of the shore, as a result the modal class for the population rose from a normal value in the mm size class, in all pre- pollution data sets, to the mm size class in April (Figure 4). In April 1997 numbers of limpets on the study shore had returned to what might be considered normal again (Figure 5) but the modal class was still shifted to the right (Figure 6). By April 1998 numbers were towards the high side of the normal range (Figure 7) and the modal class had returned to normal as well (Figure 8). Seemingly, the population had made a full recovery, in terms of numbers and size distribution, within two years: a surprisingly quick revival given the magnitude of the event. Student data sets vary in quality. However, if students are told that they will be contributing to long- term monitoring, conscientious data collection is far more likely and has occurred in this instance. Two sets of observations, a fortnight apart, by different A- level groups, in April, gave similar and statistically significant results (Archer- Thomson, 1999) (Figure 3). Having realised the value of the limpet data, monitoring continued in April of each year, to gain further insight into what might represent normal fluctuations of this population. To help with the consistency of recording, final- year MSc students from the University of Leuven, who visit FSC Dale Fort annually, were used to collect the data. In more recent years FSC Dale Fort teaching staff have collected the data. There is now an unbroken (except 5) set of data for the Frenchman s Steps population up to April 216. FIGURE 2. Frenchman s Steps and surrounding coastline. On 15 th February the Sea Empress oil tanker grounded near St Ann s Head, 2 days later high winds drove her onto the rocks. Over 7, tonnes of oil was spilt. (Map from Archer- Thomson, 1999) MATERIALS AND METHODS The site chosen was Frenchman s Steps, Grid Ref. SM82253 (Figure 2). This is a sheltered rocky shore, with a north- north- easterly aspect (Figure 1), Ballantine s Exposure Grade 4 (Ballantine, 1961). Data were collected by groups of students attending field courses at FSC Dale Fort Field Centre and latterly by FSC Dale Fort teaching staff. All data collection was directly supervised by the author. An interrupted belt transect, sampled at.75 m height intervals, was established from a fixed starting height 2.25 m Above Chart Datum (ACD). This is where the bedrock starts; below this the substrate is mobile and unsuitable habitat for limpets. Data collection continued up the shore until the upper distributional limit of limpets was reached. 5 cm by 5 cm quadrats were placed at.5 m (horizontal) intervals along a tape measure laid out at each height. In each quadrat the longest diameter of every limpet was measured and recorded in its appropriate 5 mm size class. To prevent measuring the same limpet more than once, each shell was lightly marked with chalk. The number of quadrats used each year varied with the numbers of students involved in the data collection so the results were standardised to give totals for ten quadrats at each height. No changes to the methodology have been made over the years to facilitate comparisons with past data (Archer- Thomson, 1999). Limpet population data are likely to show a degree of seasonality, especially as recruitment to the shore occurs in the autumn, so it was decided to standardise data collection to April of each year where possible; however, some data sets are from March and May. 2

3 Vertical height ACD / metres FIGURE 3. Limpet numbers at Frenchman s Steps at the given heights ACD for two April data sets and three pre- pollution examples ( ). A. April 1985 B. 29 April FIGURE 4. Limpets at Frenchman s Steps: size frequency data for a typical pre- pollution data set (A; April 1985) versus an April example (B). Modal class size is in red. 3

4 Vertical height ACD / metres FIGURE 5. Limpet numbers at Frenchman s Steps at the given heights ACD for two April data sets and three pre- pollution examples ( ), and a one- year post- pollution example (1997). A. 29 April. B. 3 April FIGURE 6. Limpets at Frenchman s Steps: size frequency data for a April data set (A) versus (B) an April 1997 (one- year post pollution). Modal class size is in red. 4

5 Vertical height ACD / metres FIGURE 7. Limpet numbers at Frenchman s Steps at the given heights ACD for two April data sets and three pre- pollution examples ( ), and two post- pollution examples (1997- one- year post pollution, 1998 two- year post pollution). A. 29 April B. 3 April 1998 FIGURE 8. Limpets at Frenchman s Steps: size frequency data for an April data set (A) versus (B) an April 1998 (two- year post pollution). Modal class size is in red. 5

6 RESULTS AND ANALYSIS Table 1 shows a data set for 3 April Table A contains class data from seven quadrats (student groups), whereas Table B gives the standardised results, as if ten quadrats had been used. Table 2 gives (standardised) results for the number of limpets found in ten 5 x 5 cm quadrats at various heights up the shore. The three data sets before were chosen because they were gathered in April and deemed typical for the shore. The two sets of data for, (1) and (2), were recorded by two different A- level groups, on the 3 and 29 of April respectively. Analysis of these results is presented in Archer- Thomson (1999). Table 3 similarly provides (standardised) data for size frequencies. TABLE 1. Original (A) and standardised (B) data for limpet numbers and size range from Frenchman s Steps on 3 April (A) data from seven quadrats (student groups). (B) the converted (standardised) data for the equivalent of ten 5 x 5 cm quadrats. The shaded cells show: the height at which most limpets were found (3.75m ACD). the mode for the size class data ( mm). the shift in the modal class at each height to progressively larger limpet sizes as height up the shore increases. (See Archer- Thomson (1999) for an explanation). A: Data from seven quadrats (student groups) Vertical height above chart datum / m TOTALS < TOTALS B: Converted (standardized) data for the equivalent of ten 5 x 5 cm quadrats Vertical height above chart datum / m TOTALS < TOTALS

7 TABLE 2. The number of limpets found in ten 5cm x 5 cm quadrats at each 75 cm vertical height interval up the shore at Frenchman s Steps, in April. ( *Data collected in March. No limpets found above 6.75 m ACD). April data Vertical height above chart datum / m Totals (1) (2) No data collected for this year * Table 4 presents the statistical analysis of the abundance data sets from Table 2. The total number of limpets at each height, for any given year, is compared with those from every other year. It should be noted that this type of data is difficult to analyse statistically. Biological data of this nature are unlikely to meet the requirements of parametric statistics (e.g. assuming normally distributed data, with equal variances) so non- parametric tests were used. Because the variation in numbers, with height up the shore was large (as a consequence of the environmental gradient on rocky shores from fully marine conditions at the bottom to near terrestrial ones at the top of the shore) the analysis had to be based on a matched- pair system (2.25 m ACD versus 2.25 m ACD for the years in question etc.). Although multivariate techniques could have been used, these would have been inaccessible to A- level students. The Wilcoxon Matched Pairs test was deemed the most suitable test. It is noted, however, that given the small sample size (seven heights maximum), this test would only record a statistically significant difference if the quadrat totals in all the heights from one year were greater (or lesser) than those from another. Inherent variation at site one (where the substrate varies from year to year) makes this constraint significant. Even so, the analysis did largely show significant differences where expected. An excellent description of the limitations associated with this statistics test is given in Fowler et al. (1998). Table 4 has been colour coded to aid interpretation. Normally, the Null Hypothesis is rejected, coded S, or accepted, coded NS, at the 5% significance level or better. In Table 4 the results have been coded (S) if the Null Hypothesis could be rejected at the 1% significance level. The decision to indicate rejection at this level is because of the inherent variability in the data (increased by substrate variation at site one, for example) and because the Wilcoxon test ideally needs a minimum sample size of six non- zero differences. Many of the data sets give six non- zero differences, some give seven, but operation here is at the lower end of the ideal sample size spectrum. Thus, (S) indicates a tendency in the data and is therefore still a useful guide. A look at Table 4 shows a cluster of significant differences (S) for the (oil pollution) data sets as expected but also another cluster of significant difference (Ss) for the 21 data and later, which highlights how different the 21, and later, population densities are. When working at the 5% significance level, there is a risk of rejecting a Null Hypothesis (H), when it is true, five times out of a hundred or one time in twenty. This is referred to as a type 1 error. Table 4 contains a great many Wilcoxon Matched Pairs test results and thus the risk of a type 1 error is increased. A Friedman test analyses the data set in table two as a whole, whilst still operating on matched pairs (It is like a matched pairs Analysis of Variance 7

8 TABLE 3. The number of limpets in each 5 mm size class on the transect up the shore at Frenchman s Steps. Data standardised for ten 5 x 5 cm quadrats at each height. Cells shaded: show the modal class. (* Data collected in March. No limpets found bigger than mm). Size class in mm April data < Totals (1) (2) No data collected this year * would be to a series of matched pair t tests, suitable if the data were normally distributed, or if sample sizes were larger). The result of the Friedman test on the data as a whole (see FRIEDMAN tab in the Excel spreadsheet of RAW data in the appendix*) allows the rejection of the H (that there is no significant difference between the medians of the data sets) at P<.1, a very highly significant result. The chance of getting a result this significant by chance is less than one in a thousand. Although the Friedman analysis reduces the chance of a type 1 error significantly, it only indicates the data sets as a whole are different, and does not show which data sets (years) are significantly different from which others. Thus, the many Wilcoxon tests are necessary despite the increased risk of type 1 errors. (* For interested readers RAW limpet survey data may be found in an additional file supplied with this on- line paper.) DISCUSSION Explanations for the patterns in the data, from 1985 to 1998, have been presented previously (Archer- Thomson, 1999) and summarised below. The total number of limpets at each height [Table 2]. Limpet densities vary with height up the shore. Numbers are low at the base of the study area, 2.25 m ACD, but increase with height to a maximum value in the middle shore (at approximately 3.75 m ACD), decreasing above this again to zero by about 6.75 m ACD. Portions of the lowest site are covered with pebbles, the amount and extent varying from year to year, which explains some of the variation at the bottom of the shore. 8

9 TABLE 4. Wilcoxon Matched Pairs Test results for limpet numbers over the given years (April data). Data standardised for ten 5 x 5 cm quadrats at each height. The shaded cells show: NS Non- significant result, Accept H (of no significant difference between the median number of limpets at the 5% significance level). S (S) Reject H at 5% significance level. Reject H at 1% significance level (see text for explanation as you would normally Accept the H here). Year (1) (2) NS NS (S) S NS NS NS NS NS NS NS NS NS S NS (S) (S) (S) (S) S NS NS NS 1986 S NS NS NS (S) S S NS S NS NS NS NS NS NS S S S S S S S 1989 (S) (S) NS NS NS NS NS NS NS NS NS S NS (S) NS NS NS NS NS NS NS (1) NS NS S S S S S NS (S) S NS S NS S S S S S S S (2) (S) S S S S S (S) S S NS (S) (S) S S S S S S S 1997 NS NS S (S) NS NS NS NS NS NS NS S S S S S (S) (S) 1998 NS NS NS NS NS S NS (S) NS NS S S S S S S NS 1999 NS NS NS NS (S) S S (S) S NS NS NS (S) NS NS NS NS NS NS S S S (S) S NS NS NS (S) NS NS NS 1 NS NS NS NS NS NS NS S S S S NS NS NS 2 NS S NS NS NS (S) (S) (S) (S) S NS NS NS 3 NS NS NS NS NS S S S S (S) S (S) 4 NS NS NS NS S S S S S S S 6 NS NS NS (S) (S) NS S S S (S) 7 NS NS (S) (S) (S) S S S S 8 NS NS NS NS S (S) NS NS 9 (S) (S) (S) S S S S 21 NS NS NS NS NS NS 211 NS NS NS NS NS 212 NS NS NS NS 213 NS NS NS 214 NS NS 215 NS 216 There is a decrease in numbers towards the top of the shore, because conditions become increasingly harsh for a marine snail. Temperature and salinity variation and dehydration create stresses that all get worse as emersion (periods spent out of water) times increase. Limpets graze selectively on green algae (both macro and microscopic), lichens and young fucoids. They often feed at night when the tide is out (Branch, 1981), or at any time when the sea is calm and the tide is in. Feeding time and food supply are not necessarily reduced in the upper shore but a limpet might feed less if already stressed by other environmental (abiotic) factors. In short, abiotic factors probably set the upper distributional limits (Branch, 1981). Numbers decrease towards the bottom of the shore because of inter- specific competition from macro algae in more sheltered regions and from barnacles in more exposed ones. Exposure to wave action will vary considerably on shores in different locations. A headland site receives far more wave energy than an embayed one, but there are subtle variations along small horizontal distances as the slope and topography vary and this can contribute over the study area width of approximately ten metres. Thus, lower distributional limits are probably set by biotic (living) factors. It follows that optimal conditions, between the two extremes mentioned above, exist for limpets in the middle shore. Over the thirty years in which data have been gathered, the trends described above have held true with no significant deviation. The total number of limpets on the shore over time [Table 2]. Looking at the population as a whole numbers were significantly reduced by the Sea Empress oil spill, recovering to normal again within a year or two. From 1998 onwards, the population fluctuated within what had become expected ( normal?) limits up until 21. In 21 and 211 limpet numbers exceeded anything hitherto 9

10 ARCHER- THOMSON (216). FIELD STUDIES ( studies- council.org/) experienced (Figure 9). Up until 21 it was assumed that normal fluctuations in the study limpet population had been established. Data from 21 and 211 threw this assumption into doubt but the question now became one of whether the record population densities were natural or if there was another explanation that might be regarded as unnatural? FIGURE 9. on Frenchman s Steps sample site, FIGURE 1. (Left) Dogwhelk Nucella lapillus plus egg capsules from Frenchman s Steps. (Right). Typical dogwhelk Nucella lapillus density in recent years (photograph taken September 9). In an attempt to explain the record high limpet densities of 21 and 211, it is necessary to include anecdotal evidence and field observations from the author s own 3- year- plus experience of the shores around FSC Dale Fort. In 1982 there were virtually no dogwhelks (Nucella lapillus) on the shores between the Field Centre and Dale village (Figure 2). Local shores might have yielded two or three animals in an area of 1 m2. Dogwhelks (Figure 1) are fairly common rocky shore animals in the UK, but at the time their numbers were severely depleted by Tri- Butyl- Tin (TBT) anti- fouling paint, which was widely used to prevent fouling organisms (barnacles, seaweeds, etc.) from growing on the hulls of pleasure boats and larger commercial vessels. In the 198s unusual changes in the sexual characteristics of dogwhelks were noticed in estuaries and other areas where small boats, painted with TBT anti- fouling paint, were concentrated (Little et al., 9). Studies showed 1 Field Studies Council (2/9/216)

11 (Gibbs et al., 1988) that female dogwhelks were growing a non- functional male reproductive organ, which blocked their oviduct, preventing successful reproduction. This condition was termed imposex. As a result, over about 2 years, many local snail populations became extinct. TBT paint was so toxic that imposex could be initiated in newly hatched snails at concentrations of only 2 ng of tin/l of seawater (Little et al., 9). Fortunately, TBT anti- fouling paints were banned from small craft (<25 m) in 1985, from all craft in 3 and all traces of the paint had to be gone from hulls by 8, by International Maritime Organisation statute. After the ban, dogwhelk populations began to recover on the shores around Dale Fort Field Centre, and they are now host to possibly unsustainably high densities (Figure 1). Dogwhelks typically eat barnacles (the acorn barnacle Semibalanus balanoides, being the favoured species (Crothers, 1985)) of which there is an abundance at the study site. A suggestion for the unusually high limpet densities in 21 and 211 is that the dogwhelks may have had such an impact on the local barnacle population that inter- specific competition for space has been reduced in the limpets favour, allowing far more to survive on the shore than hitherto. This fits in with theories about rocky shore community structure and the inter- relationships between organisms such as barnacles, limpets and dogwhelks (Jenkins et al., 1999). It is noted that after 211 (see Figure 9) the number of limpets started to decrease again. A speculative explanation can be given below, although further data will be needed to confirm if this trend back to normal continues in future years. Dogwhelks are known to also eat young limpets (Branch, 1981). In 211 there were unusually large numbers of limpets in the mm size class (Figure 11). In 212, however, when the limpet population had decreased from an all time high of 4,116 animals to 3,45, the numbers of limpets in the mm size class had decreased substantially (Figure 12). It is suggested that, having had a significant impact on the barnacle population, dogwhelks were now predating on small, thinner- shelled limpets as well, which might explain the limpet population moving towards normal again. Between 212 and 213 (Figure 13) there was a reduction in the total number of limpets on the shore (Figure 9) and a marked drop in the mm size class in particular (Figure 14). This might represent the upper size limit for limpets to be preyed on by dogwhelks (dogwhelks favour small prey if available (Crothers, 212)) but if barnacle prey is still in short supply, limpets of this size may still be a suitable option. Since 213 the total population density of limpets seems to have stabilised albeit at a higher than normal level. An interesting result from the 216 data shows an increase again in the mm size class (Figure 14). It will be useful to monitor this size class in future years to see if this represents the start of a new trend or if this is merely a short- term fluctuation. Throughout this article the terms normal, abnormal, natural and unnatural have been in inverted commas because of the difficulties in establishing whether a particular shore state is affected by human activity or not. Since the study of the Frenchman s Steps population started in the 198s and was significantly affected by Sea Empress oil (definitely unnatural ), it was assumed that the population state was normal before that pollution event. However, looking at the data again in the light of TBT effects it is entirely possible that the lack of dogwhelks in the 198s allowed a higher than previous population density of barnacles to exist and therefore the limpet densities of the time were suppressed. As dogwhelk numbers recovered after the TBT ban, barnacle densities may have been depressed allowing limpet numbers to increase to normal levels again. Thus, the recent (21 and onwards) data may not represent an unnaturally high limpet density at all but a recovery to pre- TBT levels. In summarising the trends in the above data, it is possible to recognise four different states for the limpet population on Frenchman s Steps between 1985 and 216 (Figure 9). A pre- oil pollution state, fluctuating between a total of 1, and approximately 2, individuals. An oil pollution state of approximately 1, individuals. A post- oil pollution state, but pre- 21, of 1, to 2, individuals again and a post- 21 state of 2,8 to 4, individuals. Figure 15 summarises these four states and indicates whether the differences are statistically significant, or not, with the use of 95% confidence limits as indicators. With the obvious exception of the data, it is impossible to know whether these other states indicate normal, abnormal, natural or unnatural conditions for this population. It will be fascinating to see what transpires over the next decade or so as the various dogwhelk, barnacle and limpet interactions play out. Hopefully oil pollution will not play a part in the shore dynamics again. Size frequency data [Table 3]. The modal class for 21 out of the 25 size frequency data sets is the mm one (Figures 8, 11, 12 & 13 for example). Exceptions to this rule occurred when the Sea Empress oil pollution displaced the modal class to the right ( mm size class) in and 1997 (Figures 4 & 6). There was also anomalous data in 9 for which no explanation can be given by the author. Limpets from the genus Patella grow throughout their lives with growth rates varying greatly with food supply (Branch, 1981 and Ballantine, pers. comm.). For this particular shore it is reasonable to assume that the largest limpets are the oldest and the smallest the youngest. Food supply and feeding time vary with height up the shore (Branch, 1981) so this assumption must be treated with some caution but it should hold true in general. There are fewer larger, older limpets because they die of old age, disease, etc. There are very few small, young, limpets in most of the 11

12 1 1 8 FIGURE 11. Size frequency data set, limpets of Frenchman s Steps, 19 th April 211. Modal class size is in red FIGURE 12. Size frequency data set, limpets of Frenchman s Steps, 8 th May 212. Modal class size is in red Size class / (mm) FIGURE 13. Size frequency data set, limpets of Frenchman s Steps, 11 th April 213. Modal class size is in red. 12

13 data sets. This could be for a number of reasons. Firstly, very young limpets grow through the first few size classes relatively quickly (Branch, 1981). Secondly, students collecting the data are more likely to miss the very small limpets, as they are harder to see or may even be mistaken for barnacles. Limpets settle from the plankton when their shell is approximately.2 mm long (Lewis & Bowman, 1975) and favour small, damp microhabitats such as crevices and pools. This may effectively make them invisible for sampling whilst there. The sample area at Frenchman s Steps was chosen to maximise the area of suitable rock surface and minimise the area of pools and crevices thus reducing this complication as far as possible. It is interesting to note that the population density data have varied significantly over the study period, showing four distinct phases, whereas the overall pattern in size frequency data is more consistent, varying only during, and in the two- year period after, the Sea Empress incident (9 excepting). Reasons for the shift in the modal class, during and after the oil spill are detailed in Archer- Thomson (1999). 1 1 Number of limpets FIGURE 14. Annual April shore totals ( mm size class) for limpets of Frenchman s Steps ( ). 1 Mean number of limpets at each height 8 Pre-pollution Pollution Post pollution Recent Height above chart datum / m FIGURE 15. Mean numbers of limpets, for the given heights ACD, plus 95% confidence limits summary data. 13

14 ARCHER-THOMSON (216). FIELD STUDIES ( Taxonomic note. There are three species of limpet in the genus Patella that occur on rocky shores in south-west Britain, Patella vulgata Linnaeus, common limpet; Patella ulyssiponensis Gmelin, china limpet; and Patella intermedia Murray in Knapp, black-footed limpet. An excellent guide to their identification and biology is given in Fish & Fish (211). Identification of the three species in the field with student groups is, to all practical purposes, impossible. It involves taking limpets off the rock to look at their foot colour and this invariably kills them. This would be unacceptably destructive on a site where long-term monitoring is intended and, more importantly, ethically unacceptable, especially with groups of students who are being taught to respect the environment in which they study. Fortunately P. vulgata is by far the most common limpet on the shore at Frenchman s Steps (Ballantine pers. com. and author s own experience), the other two species favour more wave exposed locations and the lower part of the shore. Patella ulyssiponensis also favours rock pools in the middle shore, where it feeds on encrusting red algae, but again there are no pools at the chosen sample heights. Therefore although there may be representatives of the other two species of limpet in the study area, it is likely their abundance is low and so will not affect the results significantly. Biological note. Limpets have recently made the news (bbc.co.uk, science & environment, 18 th Feb 215) as they have the honour of hosting the strongest biological material ever tested having ousted spider silk from the top spot (Barber et al., 215). Limpets feed using a tongue-like structure called a radula, which they scrape over the rock surface as they feed on various species of green seaweed, young stages of brown seaweeds, microscopic algae and other components of what is referred to as biofilm covering the rock surface. The radula is essentially a ribbon covered in teeth that are less than a millimetre in length (see excellent high magnification images of a limpet radula in Cremona, 214, page 21). The teeth are made up of fibres consisting of an iron-based mineral called goethite laced through a protein base. Because the mineral fibres are so thin they are not prone to structural flaws that would normally weaken materials. The remarkable feature of these fibres is that the extreme strength is scale-independent, normally structures, like bridges and arches for example, have to be of a certain size to achieve their desired tensile strength; not so limpet radula teeth. The material is so strong that it can be compared to a single string of spaghetti holding up 1, one-kilogram bags of sugar! China limpets exploit the strength of their radula teeth to eat various species of pink paint weed in turn the paint weed has its growing point well below the surface so, even with a growth rate of less than a millimetre a year, it can withstand limpet grazing. CONCLUDING REMARKS Numbers of limpets on the shore at Frenchman s Steps show considerable variation over the years that they have been studied by Dale Fort students and educational staff. Four main states have been identified namely, pre-oil pollution ( data sets), pollution and immediate aftermath (-1997 data sets), post-oil pollution ( data sets) and post 21 ( data sets). With the obvious exception of the oil pollution results assigning normal/ abnormal status to any of the data sets is problematic because of the plethora of other variables involved. Size frequency data show a remarkable consistency across the years with the norm only being disrupted by a major oil spill and then for only two years. Simple fieldwork, carried out with student groups armed with nothing more complex than quadrats, calipers or 15 cm rules, chalk, recording sheets, optical levels and metre rules can yield fascinating data. These results, however, reveal patterns that can be difficult to interpret with any certainty because of the number of variables acting on the shore community. This includes supply-side considerations to do with recruitment to the population and planktonic phases of shore organisms. One of the many features that keep the monitoring of this shore exciting is that the interpretation of the patterns keeps changing as each new data set is collected. Roll on the next thirty years! DEDICATION AND THANKS I would like to dedicate this paper to the memory of Dr. Bill Ballantine M.B.E. who knew more about limpets than most, was a staunch supporter of FSC Dale Fort from the 195s and beyond, and an inspiration for the cause of marine conservation worldwide, latterly from his base in New Zealand. Hopefully he still investigates limpet biology, albeit on a distant shore. I would also like to thank the various student groups, especially those from the University of Leuven, and the education staff of FSC Dale Fort for their help in gathering these data over the years. It would have been spectacularly tedious without you! REFERENCES Archer-Thomson, J.H.S., (1999). The Sea Empress incident and the limpets of Frenchman s Steps. Field Studies, 9, Ballantine, W.J., (1961). A biologically-defined exposure scale for the comparative description of rocky shores. Field Studies, 1 (3), 1-19.

15 Barber, J.H., Lu, D. & Pugno, N.M., (215). Extreme strength observed in limpet teeth. Journal of the Royal Society Interface, 12: Branch, G.M., (1981). The biology of limpets: physical factors, energy flow and ecological interactions. Oceanography and Marine Biology: an Annual Review, 19, Crothers, J.H., (1985). Dog- whelks: an introduction to the biology of Nucella lapillus (L.). Field Studies, 6, Crothers, J.H., (212). Snails on rocky sea shores. Naturalists Handbook 3. Pelagic Publishing, Exeter. Cremona, J., (214). Seashores: an ecological guide. Crowood Books, Wiltshire. Fish, J.D. & Fish, S., (211). A student s guide to the seashore, 3 rd edn. Cambridge University Press, Cambridge. Crump, R. G., Morley, H.S. & Williams, A.D., (1998). West Angle Bay, a case study. Littoral monitoring of permanent quadrats before and after the Sea Empress oil spill. Field Studies, 9, Fowler, J., Cohen, L. & Jarvis, P., (1998). Practical Statistics for Field Biology, 2 nd edn. Wiley, England. Gibbs, P.E., Pascoe, P.L., & Burt, G.R. (1988). Sex change in the female dog- whelk, Nucella lapillus, induced by tributyltin from antifouling paints. Journal of the Marine Biological Association of the UK, 68, Jenkins, S.R., Hawkin, S.J., & Norton, T.A. (1999). Direct and indirect effects of a macroalgal canopy and limpet grazing in structuring a sheltered intertidal community. Marine Ecology Progress Series, 188, Lewis, J.R. & Bowman, R.S., (1975). Local habitat induced variations in the population dynamics of Patella vulgata. Journal of Experimental Marine Biology and Ecology, 17, Little, C., Williams, G.A. & Trowbridge, C.D., (9). The Biology of Rocky Shores, 2 nd edn. Biology of Habitats. Oxford University Press, Oxford. Link for raw data FURTHER INFORMATION 15