Meliora Submission

Introduction

Humans have been impacting shallow lakes around the world for thousands of years, with nutrient enrichment impacting a large number of the lakes globally (Scheffer, 2004). Despite this, the effect eutrophication has on shallow lakes, both past and present, are poorly documented and understood. The lack of long–term, reliable data on many lakes (Smol and Stoermer, 2010) means studies are limited in understanding of the past ecological responses. One of the earlier anthropogenic impacts on shallow lakes were the building of medieval settlements, known as crannogs. Crannogs are also understudied, with many studies lacking modern stratigraphic controls due to early excavation (Selby and Brown, 2007). Crannog studies have the potential to improve knowledge on the eutrophication and ecosystem response. As crannogs were built and abandoned thousands of years ago, the ecology during the disturbance, as well as before and after, can be investigated, something that cannot be achieved in modern eutrophication due to how recent the changes have occurred.

There has been a renewed academic interest in both shallow lakes and crannogs in the last two decades because of advances in palaeoecology, in particular the use of diatoms. Diatoms can be used to infer past trophic states and ecological conditions of shallow lakes (Bradshaw, 2002). Their abundance and sensitivity have resulted in their widespread use to reconstruct the past environments of lakes as different species have different environmental conditions. To improve the understanding of the relationship between shallow lakes and eutrophication, both past and present, further study is required and diatoms provide an accurate way of reconstructing the ecosystem response to nutrient enrichment. Though there are numerous models and theories that can be used to explain the ecosystem response and recovery of shallow lakes, they are not often replicated in reality (Scheffer 2004; Battarbee et al., 2012). Shallow lakes are incredibly complex ecosystems and depending on the level of eutrophication, each will respond differently. This study sets out to improve the understanding of the ecological response and recovery of shallow lakes when subject to nutrient enrichment.

Research Questions and Aims

Research question: What impact did a medieval crannog have on the ecosystem of White Loch of Myrton and has the ecosystem recovered?

The aims of this dissertation are as follows;

  • Reconstruct the ecological effects the crannog phase may have had on the White Loch of Myrton using diatom analysis.

–Any changes in the diatom assemblage will be evident in the diatom stratigraphy and compared to any changes in total phosphorus and species diversity.

–A detrended correspondence analysis (DCA) will be undertaken to identify the main environmental gradients that have caused the variation in the diatom assemblage.

  • Analysis the recovery pathway of the lake ecosystem post–crannog and whether the lake has fully recovered

–A phase space diagram will be constructed to show the deterioration and recovery trajectory to analyse whether the lake has recovered.

Literature Review

Palaeolimnology

Palaeolimnology is the study of past lakes and the relationships between their chemical, biological and physical features (Birks and Birks, 1980). Lake sediment is often collected by vertical coring where various laboratory procedures can be undertaken (Lowe and Walker, 2015) to gain information about paleoclimate change and ecosystem development. As many sediment cores are fossilerous, palaeolimnology is interlinked with palaeoecology; the relationship between past organisms and the environment in which they lived. Microfossils are the main source of information about past lake environments (Birks and Birks, 1980).

Microfossils are useful as each species within a community has a different environmental tolerance over various environmental gradients; a niche. If the ecosystem requirements and tolerances of the species within a community are known, as well as relative species abundance, microfossils can be used as a proxy for palaeoclimate and ecosystem change (Boulangeat et al., 2012). Combining palaeolimnology and palaeoecology can result in a multiproxy study which are more continuous and provide a more sophisticated reconstruction of lake environments.

Shallow Lakes

Despite efforts there is no official definition for shallow lakes, though it is generally agreed that the depth does not exceed 3m (Scheffer, 2004). Shallow lakes are always polymictic so lack stratification of oxygen or temperature. Constant mixing of water allows sediment and nutrients to settle before reentering the water column due to resuspension and the lack of thermocline (Scheffer, 2004). This means the pelagic zone does not have a loss of nutrients in the Summer months. The resuspension and lack of stratification makes the processes and dynamics within shallow lakes considerably different to deep lakes.

Whilst the differences between the two lake types are acknowledged, there are fewer studies on shallow lakes. Interest has increased in the past 20 years, and in doing so, has highlighted their importance on a global scale. They are one of the most threatened ecosystems with evidence of decreasing biodiversity (Davidson et al.,2013); a concern for a habitat with high species diversity and richness. There is a need to improve knowledge on the causes, rates and timings of change in shallow lakes (Smol and Stoermer, 2010), however, most studies that investigate the anthropogenic impacts on lake systems are on deep lakes (Selby and Brown, 2007).

Ecosystem Shift and Recovery Shifts

Shifts in ecosystems occur when conditions, internal or external, are changed. The way in which an ecosystem changes varies and can be; linear, threshold dependant non–linear or threshold dependant critical transition (Figure 1). The latter, unlike the two former, is irreversible due to a tipping point being met and changes occurring abruptly. Conditions do not have to change drastically to create large shifts in a system (Scheffer and Carpenter, 2003).

Eutrophication is the nutrient enrichment of water, in particular phosphorus, and is a common cause shallow lake ecosystem shifts. The recovery of the ecosystem, after the nutrient input (external loading) has returned to normal levels, is complex. This is due to a process called internal loading where phosphorus is slowly released by sediment and dead organisms, which stored phosphorus during its excess. As the sediment water column ratio is high in shallow lakes, the sediment can affect the nutrient concentration (Sondergaard et al., 2003). This causes delays in shallow lake recovery and can take several years (Bennion et al., 2015). Eutrophication is the main anthropogenic impact on shallow lakes and has been responsible for the decrease in the quality of shallow lakes over the last past century (Moss, 2010).

Recovery pathways are difficult to follow due to the low resolution of sediments, resulting in difficulties pinpointing exactly when nutrient levels increased and decreased or calculate the rates of change (Randsalu–Wendrup et al., 2016). A holistic approach, which looks at past and present changes in nutrients, is required to further understanding (Battarbee et al., 2005). The natural recovery of lakes needs to be understood so it is known when biomanipulation has been successful.

Figure 1The various ways in which an ecosystem can respond to changes in conditions. Scheffer et al., (2001)

Diatoms

Diatoms are unicellular photosynthetic algae belonging to the class Bacillariophyceae. They are found in most aquatic environments at a range of temperature, nutrient and pH regimes. They are formed from two valves which are identical in shape but slightly different in size. Classification of diatoms is based around the characteristics of the valves (Lowe et al., 2004). Diatoms can be divided into Centrales taxa where valves form around a point, or Pennate taxa (Figure 2) where valves form along a plane.

Figure 2. Batterbee et al., 2001) Examples of Centrale diatom(left) and a Pennate diatom (right) with its raphe running in the middle

Diatoms can be used to reconstruct changes occurring in fluvial, lake, marine and extreme environments. The use of diatoms in paleolimnology has increased since the 1950s due to the 'age of objectification' whereby diatom growth and abundance can be linked to variables in a precise, quantitate manner. The four key selection pressures to benthic diatoms are; light, nutrients, grazing and physical disturbance (Lowe, 1996) making diatoms particularly ecologically sensitive.

Diatoms can be used to answer ecological questions and have the potential to be powerful indicators to environmental change. Their short generation times leads to quick responses to changes in water quality relative to macrophytes and invertebrates (McCormick and Stevenson, 1998). This has led to diatoms being an important early signal to eutrophication as well as other water quality changes. environmental investigations.

In shallow lakes specifically, diatoms have a high abundance and diversity due to a larger littoral zone relative to the pelagic zone. The constant mixing within shallow lakes also increases diatom abundance. Benthic diatoms can contribute between 1% and 100% to shallow lake production (Vadeboncoeur et al., 2003) and is a function of the lakes trophic state.

Crannogs

Crannogs are partially or wholly artificial islands that are found in shallow lakes (Barbe and Crone, 1993). The island is usually connected by a causeway made from stone or timber with the average diameter measuring 25m (Fredengren, 2002). Crannogs are found in Scotland, Ireland and one in South Wales. They date back as far as the Iron Age and Medieval Age (Fredengren, 2002). Scottish crannogs date to the Neolithic period and are older than Irish crannogs (Cavers, 2006) although there is no known explanation for why. There are numerous ideas as to what crannogs were used for, with palynology evidence suggesting; cereal cultivation, animal domestication and metal work (Cavers, 2006).

As crannogs are not fully understood, with numerous unanswered questions, there is a need for further study. This is particularly true of Scottish crannogs, which has seen fewer comprehensive studies undertaken than Irish crannogs. The numerous gaps in crannog knowledge can only be filled with further study, especially in Scotland.

Transfer Function

Reconstructing eutrophication trends over the last 20 years have improved, with estimates now being more accurate and easier to predict (Smol and Stoermer, 2010). Prior to recent developments, eutrophication reconstructions were qualitative, using shifts in diatom species as proof of trophic status change. Due to transfer function, quantitative estimates can be made as relationships between diatoms and total phosphorus (TP) have been established. This is done by linking modern diatom assemblages found in sediment surface to the known environmental conditions of the lake. This is repeated on numerous lakes with varying environmental conditions and then combined to form a training set, which is then compared to fossil diatom assemblages.

There are issues with the technique as well as some unrealistic assumptions. It is assumed that the modern diatom species have the same ecological requirements and responses as their fossil counterparts and have not changed over time (Smol and Stoermer, 2010). The function does not take in to account the ecological complexities of diatoms, with phosphorus being the only factor (Sayer, 2001). Both of these assumptions are likely to be true. There is also an issue with the 'edge effect' where low TP levels are overestimated and high TP levels underestimated due to poor training set quality (Sayer, 2001).

Despite numerous limitations to the transfer function, without it, reconstruction of TP levels in lakes would be incredibly difficult due to the lack of long term water chemistry data. Training set data is constantly improving, meaning TP predictions are becoming more realistic, as training sets are being combined to reduce the edge effect. Often, inferred and observed TP are similar in the majority of lakes (Anderson et al., 1993; Sayer 2001). DI– TP is a well–developed and widely used technique that can help with the understanding of nutrient enriched lakes, where other techniques cannot.

Study Site

White Loch of Myrton is a shallow lake (Figure 3) situated in Wigtownshire, South West Scotland (54.757405, –4.560562). The lake is eutrophic and relatively large with an area of 20Ha, but has a maximum depth of only 12m (Crone and Cavers 2016). Wigtownshire has other, smaller lochs in the county, including the shrunken Black Loch of Myrton. In 1885, when the crannog was revealed Wigtownshire was exclusively an agricultural and grazing county (Groome, 1885) with these sectors remaining significant today.

Figure 3. The location of the White Loch of Myrton is marked on the map with an X, and is situated near Port William. The shape, size and surrounding of the lake is seen on the right

A series of cores of the loch were taken in 2012 which were highly organic, peaty and consisted of animal bones, charcoal and wood. In the upper depths there were remains of piles and at lower depths paving. The difference in time between these structures suggest the crannog was used to a later date than other studied crannogs. The core has been radiocarbon dated at 2080±50 yr BP ( 350 BC to AD 50). The crannog itself had trees growing on it, with reeds surrounding it on the south and west sides of the island (Henderson et al., 2003). The north and east were surrounded by piles.

White Loch of Myrton was chosen to study as it is naturally eutrophic nature may alter the results that are expected if nutrient rich diatoms are already present. The possible late use of the crannog may also give different results compared to other crannogs and tipping points may be reached. Finally, the Black Loch of Myrton, which was situated nearby, no longer has open water and new drainage may pose a threat to the organic remains of the crannog site (Cavers, 2010). Studying White Loch of Myrton may aid findings or fill missing gaps in the Black Loch of Myrton crannog study.

Methods

Core excavation

Cores were excavated in April 2016 12m away from the crannog centre and at a water depth of 1.6m. Cores sections overlapped to reduce the chances of an incomplete core (Fonvile, 2015). The cores were stored in guttering and wrapped in cling film to avoid damage. On arrival the Southampton, they were stored at temperatures < 8°C.

Diatom sampling and preparation

Sixteen subsamples were taken at a range of intervals, depending on how the core related to the crannog history. It was clear within the stratigraphy where the crannog had occurred in relation to the core due to large wood fragments. Subsamples were taken at 16–18cm intervals in areas where it was visible that the crannog had existed. At post and pre–crannog points along the core, subsamples were taken at 8–10cm. There was a higher resolution at these points so transition zones can be analysed. After preliminary data analysis had been undertaken, the decision was made to take two more samples at 70.5cm and 130.5cm, to increase the resolution of these transition zones further.

Chemical digestion by hydrogen peroxide was undertaken to prepare samples following Battarbee (1986) This removes any material that may hinder diatom analysis. Slide preparation was finalised following Renberg's (1990) procedure with the mounting of samples on to microscopic slides using Naprax. A minimum of 250 diatom valves were counted and identified per slide at x100 oil immersion objective using, predominately, a Nikon Eclipse 80i microscope with phase contrast. Krammer and Lange–Bertalot (1999a, 1990b, 2000 and 2004) were used to identify valves.

Data Analysis

Once all raw data was collected, it was inputted into Excel and each diatom taxa was classified in to either tychoplanktonic, planktonic or periphytic, based on groupings by Matton (2016). Creating classifications made it easier to see which groups were the most abundant at each depth. Two diatoms that were particularly abundant at various depths in the core were removed from their classification and formed individual groups. Although Aulacoseira granulata and Cyclostephanos dubius are periphytic and tychoplanktonic, respectively, their abundance changes appeared to differ to that of their main groups.

Tilia

A diatom stratigraphy diagram was produced using Tilia (Grimm, 2011), which manages palaeontology data. Due to the large number of diatom taxa counted, data was formatted prior to inputting into Tilia. It is standard to class any diatom with a percentage occurrence of <2% as 'rare.' However, the diatom count at 2% remained high at 66 species. It was decided that for the stratigraphy diagram, diatoms with a percentage count of <6% were classed as 'rare' and reduced the diatom assemblage from 143 to 33. With a smaller number of taxa it was possible to display data in a stratigraphy graph. Rare diatoms were combined into one column in Tilia, so their response to changes are still represented.

PAST

Cluster analysis and DCA were performed in PAST (Hammer et al., 2001). A DCA is one of the most widely used techniques of ordination. As species percentages were not normally distributed a ssquare root transformation was adopted as it avoids skewing. Diatoms that had an abundance of < 2% were not used in the DCA to avoid any unwanted influence.

Cluster analysis creates a dendrogram by creating hierarchical groupings, with a greater distance between depth showing a greater dissimilarity. Depths were kept in order by keeping them constrained so major changes can be visibly identified. Euclidean distance was used to measure similarity distance as it is he most commonly used measure, using the simple straight line distance (Hair et al., 2010).

Species diversity was measured using the Shannon Index. Whilst there are a number of diversity indices, the Shannon Index was chosen as it is most sensitive to species of medium importance (Peet, 1974). It is one of the best indices to show the changes of rarer species and is why Simpson Diversity was not chosen. No transformation or removal as species was required before measuring diversity.

Square chord distance (SCD) was used to measure the degree of floristic change within the diatom assemblages (Bennion et al., 2015) between the deepest sample depth (197.5cm) and the other depths. Species count data, with no species being removed or transformed so the full assemblages could be compared to the reference assemblage. Whilst 197.5cm was the deepest sample depth, preliminary analysis showed it differed greatly to the other pre–crannog depths. Instead 177.5cm was used as the reference depth. Significance was tested to the fifth percentile, which deems any SCD score with a value of <0.48 as having an insignificant floristic change.

Transfer Function

The transfer function was undertaken using the statistical computing software R (R, 2015) and used the European Diatom Database (EDDI,1998 ) to download various training sets. North West Europe, Southern England, Northern Ireland and the Combined dataset were downloaded as the training data in these sets were most likely to have similar assemblages to the White Loch of Myrton fossil dataset. Prior to using R, each diatom name was replaced with its relevant code name from EDDI to allow comparison of datasets. Not all codes were available, with only 85 of the 143 diatoms in the fossil dataset being used in the transfer function.

In R, the fossil dataset was individually compared to the four training datasets using a Conical Correlation Analysis (CCA) by merging the data sets together. The WA–PLS TP was adopted over the WA regression as WAPLS reduces edge effect and uses patterns in the data to update the TP which reduces error (Smol and Stoermer, 2010). After visual analysis, cross correlation between fossil datasets and training sets was undertaken. The Northwest European set was used as it had highest R2 value (0.723) and lowest RMSE value (0.269).

Results

Cluster Analysis

The dendrogram appears to show 3 distinct zones (Figure 4), with the third zone showing subzones. When compared to Ti concentrations (which signifies clay and erosion from the crannog) the crannog boundaries suggested by cluster analysis coincide with Ti peaks.

Cluster analysis was compared to core stratigraphy and Ti concentration to see if there were similarities. The core stratigraphy shows evidence of a crannog at 130–40cm, which is slightly longer than the cluster analysis shows. When compared to the Ti concentrations (which signifies erosion and clay from the crannog), the Ti peak occurs at the crannog boundaries suggested by the cluster analysis.

Figure 4 Dendrogram showing the main three zones of the core, with the proposed crannog phase in the highlighted area

Diatom Stratigraphy

The diatom assemblage can be seen in Figure 5, where the core is dominated by tychoplanktonic taxa, with 54% of the assemblage fitting this group, when A.granulata and C.dubius are included. Periphytic and planktonic taxa are also present at similar percentage to each other . Pseudostavrosia brevistrata, A.granulata and C.dubius, which are all tychoplanktonic, are the 3 most dominate diatom throughout the core, though the Flagilaria spp. also has high abundances. The stratigraphy was divided using the zones produced from the cluster analysis.

WLM–1 (197.5–138.5cm)

This is at the bottom of the core, and is dominated by Flagilaria spp., especially Flagilaria psuedoconstruens, though the percentage decreases rapidly at the top of this zone. Other tychoplanktonic species are present at high percentages with fluctuations occurring within the core. Periphytic and planktonic are present at 22.12% and 27.94% respectively, with the former taxa decreasing with distance to the top of the zone. The TP shows an overall increasing trend whilst LOI550 is constant until 130cm where there is a sharp rise. Both Si and Ti have low but constant values.

WLM–2 (130.5–84.5cm)

In this zone A.granulata increases dramatically making up 31.6% of the taxa, whilst C.dubius also increases to 19.9%. This zone is dominated heavily by these two species. Every other diatom speciies present, except those that have only emerged in this zone, have decreased. with very few present. The diatoms that were not present in WLM–1 are all centric diatoms. All taxa groups decrease to <17%. TP and LOI550 rise in this zone, though the latter shows some fluctuations. Ti and Si also increase, with peaks at 104cm and 70cm respectively, with Si showing a sharp increase.

WLM–3 (70.5–4.5cm)

In this zone periphytic taxa are the most abundant, with a percentage over twice as high as WLM–1 and WLM–3, mainly caused by a large increase in Cocconeis placentula. A.granulata decreases to the lowest seen in all the zones, whilst C.dubius decreases to a lesser extent. The percentage of tychoplanktonic taxa increases from WLM–2, but not to the levels of WLM–1 due to the Flagilaria spp. only increasing by small amounts. TP gradually decreases until 28.5cm before a sharp increase, a trend similar to LOI550. There is a sharp decline in Si at the start of this zone and Ti remains constant.

Figure 5 Diatom stratigraphy with the proposed crannog phase between the red lines

Shannon Index

At the bottom of the core, diversity is high with values ranging from 2.96–3.02. A decrease in diversity occurs at 130.5cm, which is at the start of the proposed crannog phase. Diversity then rapidly increase, reaching a diversity of 3.17 at 64.5cm, which is post–crannog. There is then a very gradual decrease in diversity until the top of the core. There are large fluctuations of diversity within the crannog phase. TP and diversity have with a negative correlation (Figure 6) of –1 (p>0.001) when Pearsons correlation statistics were computed between TP and diversity.

Figure 6 . Trends of TP (red) and diversity (black) along the core

DCA results

The eigenvalues for DCA1 and DCA2 are 0.2974 and 0.2027 respectively, and account for 50% of all the variance within the diatom assemblage. At the bottom of the core, values are steady, with little change and DCA2 values are higher than DCA1 (Figure 7). At 138.5cm, values increase for both DCA1 and DCA2, though DCA1 increases rapidly, with its values now higher than those of DCA2, despite DCA2 showing its first of two peaks at a value of 130. Both DCAs show a decrease, though DCA2 decreases rapidly to a value of 7 before peaking at 70.5cm with a value of 191. Both DCAs then decrease, with DCA1 values decrease being short lived as an increase occurs at 44.5 which continues to the top of the core. DCA2s decrease is longer and more pronounced, with a value of 0 occurring at 125cm before a steady increase. Prior to the crannog phase, DCA2 had higher values, but this changes post–crannog with DCA1 having higher values.

Figure 7 Relationship between TP (red), DCA1 (green) and DCA2 (blue) . The crannog phase is within the highlighted area.

DCA1, DCA2 and TP

TP was compared to both DCA1 and DCA2, due the similarity of the eigenvalues (Figure 7). TP and DCA1 have a similar overall trend, although there are differences along the core. The general trend for both TP an DCA1 are low values at the bottom of the core, followed by a sharp rise in values before a decrease and finally increasing in value near the top of the core to high values. There are more fluctuations in the TP, than DCA1 at the bottom of the core, however, the first increase occurs at similar depths. The TP increase begins at 146.5cm and DCA1 begins at 154.5cm. The TP peak, which is likely to signify crannog usage, is markedly shorter than the DCA1 peak; only lasting from 122.5–84.5cm, rather than 138.5–70.5cm. The TP and DCA1 also decrease at different points in the core. DCA1, decreases until 44.5 before a rapid increase, whilst TP shows sharp decline until 28.5. Both have higher values in the top of the core than any other part of the core, including the crannog phase. The lowest TP and DCA1 values occur at 177.5 and 146.5, meanwhile the highest values both occur at 4.5cm. Due to the similarity between TP and DCA1 trends, the relationship between TP and DCA2 are similar to that of DCA1 and DCA2.

Phase Space

The phase space of White Loch of Myrton is very complex and does not follow a linear path (Figure 8). The bottom of the core until 146.5cm is in a space with low TP and DCA1 values, with the depths closely clustered together. There is then a sharp increase in TP occurs at 138,5cm, though DCA1 remains low. At 130.5cm, TP decreases slightly and DCA1 increases rapidly into a different space. Between 122.5–84.5, the plot moves to the top right of the space with high TP and DCA1 values. The plot then moves left at 130.5cm, back to a similar space as 70.5cm. The DCA values decrease and there is a clustering of depths between 64.5–28.5cm before an increase in TP and DCA1 values, seeing the plot move in to the top right land corner of the space.

Figure 8Phase space diagram, where numbers indicate the depths

Square Chord

The SCD dissimilarity scores are from 0–2 where 0 means the assemblages are exactly the same and 2 means they are completely different. There is significant change at the fifth percentile between 177.5cm and all the other depths, though changes in the crannog phases are greater (Figure 9) where there are high levels of dissimilarity with scores of >1.

Figure 9. Dissimilarity scores showing the degree of floristic change with the crannog phase in the highlighted area

Discussion

Ecological impacts

The varying preferences of diatom species can be used to link changes seen in the diatom assemblage to changes in the lake conditions. The abundance of the Flagilaria spp. in WLM–1 can be attributed to the White Loch of Myrton being a clear and mesotrophic, as this is where the Flagilaria spp. thrives (Sayer, 2001). Meanwhile, the increase of C.dubius at the transition zone between WLM–1 and WLM–2 signifies there is early nutrient enrichment. C.dubius is often one of the first diatom species to respond to an increase in nutrient input (Sayer, 2001) and is likely to signify the start of crannog usage. C.dubius and A.granulata are dominant in WLM–2, though they are abundance throughout, possibly because of the White Loch of Myrton's naturally eutrophic natures. C.dubius and A.granulata do signify eutrophication though their cosmopolitan nature means they can thrive in many environmental conditions. The increase in Stephanodiscus spp.. also signifies eutrophication (Witkowski and Pempkowick, 1995) and an in–wash of humic substances, suggesting crannog usage. It would be expected that in WLM–2 the lake conditions would favour planktonic species, as the increase in nutrients causes algae bloom, blocking out sunlight required by periphytic diatoms (Selby and Brown, 2007). This is not the case, with periphytic diatoms being the most dominate of the three groups. The crannog may have provided extra habitats for the periphytic diatoms due to sediment accumulation and debris (Bradbury, 1975). The ecological impacts of nutrient enrichment from the crannog may differ from modern eutrophication, due to how intrusive the crannog structure is on the lake. Modern eutrophication rarely involves a structure being added into the lake, therefore there is no habitat change. Another explanation for the lack of planktonic species in WLM–2 is the lack of Si. Si is a limiting factor to the growth of planktonic species, more than other groups of diatoms (Peinerud, 1997). The LOI550 also shows that there is an increase in organic matter at WLM–2, most likely waste products from the crannog. An increase in organic matter can cause an increase in numerous processes that further change the ecosystem (Dean, 2006), though these changes are unlikely to be shown in the diatom assemblage as organic matter is not the main driver.

The recovery of diatom assemblages can take as little as 5 years (Little et al., 2000), so it could be assumed that the WLM–3 assemblage would show similarities with WLM–1. Often this is not the case with post and pre disturbance assemblages differing, though have ecologically equivalent diatom species. This is seen at the White Loch of Myrton with WLM–1 and WLM–3 having ecologically similar assemblages. There is the recurrence of the Flagilaria spp. and the cosmopolitan species return to post–disturbance levels. Despite this, there are some key differences between the assemblages of WLM–1 and WLM–3. The Stephanodiscus spp.., which is symbolic of eutrophication, remains present post–disturbance, though it has been noticed that some planktonic species do remain at a high abundance despite the end of the disturbance (Bennion et al., 2015). There is a large peak in Si content at the boundary of WLM–2 and WLM3, which could suggest an algae bloom (Hurley et al., 1985). This idea is supported by the peak in the number of rare species.

There are key ecological differences within the 3 zones, with the crannog decreasing stability and ecological quality. Though there are some similarities between WLM–1 and WLM–3, the crannog has caused long term ecological effects within the lake. Species remain that suit eutrophic conditions and there is a high percentage of rare species due to the changes within the lake conditions. Whilst there has been some return back to the original assemblage the crannog appears to have had long term impacts on the ecology of the lake, despite sufficient time to recover.

Species diversity and rarity

Species diversity can be used as a useful indicator for the wellbeing of an ecological system (Magurran, 1988), with greater diversity signifying a more stable ecosystem. A decrease in diversity is linked to pollution and nutrient excess, with some the diversity of some nutrient enriched lakes decreasing by half (Toporowska et al., 2010). Between 130.5 and 70.5cm, where diversity is low, the ecological quality could be described as poor. The diversity decrease is likely to have been caused by nutrients from the crannog being disposed of in the lake. The continuous addition of nutrients favours competitive diatom species, resulting in certain species ( A.granulata and C.dubius) dominating (Dobson and Field 1998). The small increase in diversity at 104.5cm within the enriched phase of the core, may signify optimum usage of the crannog. A sudden pulse in nutrients, as opposed to continuous addition, allows opportunistic species to coexist with the more dominate species (Dobson and Field, 1998). This, however, is short lived as opportunistic species abundance crashes shortly after the pulse. Although the correlation between TP and diversity is very negatively significant at –1.0, there is also a significant negative correlation between DCA2 and diversity (p<0.005). This suggests that, whilst TP is the main driver of diversity changes, the environmental gradient for DCA2 is also having ecological effects.

The percentage of rare species is constant in WLM–1, but increases in the other two zones. The increase in species richness has a positive relationship with disturbances (Kondoh, 2001), as favourable events and conditions allow opportunistic species to thrive (Levinton, 1970). The percentage of rare species remains high in WLM–3, suggesting disturbances are still occurring and that conditions are not stable within the ecosystem

There are issues with using WLM–1 as the reference condition of the lake as there is no way to be certain it is in its natural state. In a study by Bigler et al., (2007), A.granulata is not present, except for in eutrophic lakes. However it is present throughout the diatom assemblage of the White Loch of Myrton. Whilst the White Loch of Myrton is naturally eutrophic, the abundance of A.granulata is similar in WLM–1 and WLM–3, despite WLM–3 still recovering from the crannog. It could therefore be argued that there has been some human impact on the lake prior to the crannog phase. In support of this idea, there is evidence that just 17 miles north of White Loch of Myrton, there was a large village dating –4500 yr BP (University of Manchester, No Date). Meanwhile, the charcoal in WLM–2 has been dated at 2245±29 yr BP. It is likely that in this 2000 year period, humans would have used White Loch of Myrton, as it is one of the largest lakes in the area. This would mean that WLM–1 is not a reflection of the lakes natural state, making comparison difficult.

DCA1 and DCA2

The diatom data is multidimensional, with a number of variables causing the variability within the species composition (Palmer 1993), though only a few are required by ecologists. Often only 13 dimensions are required to explain the majority of the data variability. A DCA arranges species along an unknown environmental gradient and the species composition can be used to infer what the environmental gradient is. Together DCA1 and DCA2 explained 50% of the variability in the data, where DCA1 is the main driver. Diatom species are determined by numerous physical and chemical parameters (Smol and Stroermer, 2010), so the species with the most extreme axis values are compared. In DCA1, Cyclotella comta had the lowest value whilst C.pediculus and Stephanodiscus parvus had the highest values. C.comta is usually found in oligtrophic or mesotrophic lakes (Huber et al., 2007), whilst C.pediculus and S.parvus are found in nutrient rich lakes and are tolerant to pollution (Sayer, 2001). The environmental gradient for DCA1 is most likely nutrient levels. This is supported by A.granulata and S.minutulus being in the middle of the gradient. Although these two species were dominant in the nutrient rich crannog phase, they are cosmopolitan species (Lobo et al., 2010.) A Pearson correlation between TP and DCA1 was undertaken to see if the phosphorus was the nutrient causing the change. The result was a significant positive correlation of 0.612 (p<0.001). Although the correlation is not very strong, it is statistically significant and TP likely to be the environmental gradient of DCA1.

The environmental gradient for DCA2, again, could be a range of factors although the most likely is pH. Tabellaria flocculosa and Navicula radiosa have the highest DCA2 values and both prefer alkaline waters. Navicula cryptotenella, which has the lowest DCA2 value, prefers acidic waters (Fonville, 2015). Changes to the pH of water can be manmade or natural, though the cause in this lake is likely to be the former. Waste products from the crannog or an increase in photosynthesis products may have made the lake more basic. However, pH is affected by TP, with an increase in TP leading to an increase in alkalinity (Lotter et al., 1998). If DCA2 is pH, it would be expected that there would be a correlation between DCA1 and DCA2. Although statistically insignificant, there was no correlation (–0.193) between the two DCA's. Whilst the diatom assemblage may suggest pH to be the environmental gradient of DCA2, it cannot be statistically proved. The unknown DCA2 driver has a negative correlation with diversity, which could be used in the future to help decipher the DCA2 driver.

The TP results throughout the crannog are realistically too high to be accepted as the true TP values. A.granulata's TP maxima is 71µg/L though occurs at 274µg/L and F.pseudocontruen occurs at 217µg/L whilst having a TP maxima of <4µg/L (Bigler et al., 2007) Although the combined North West Europe training set was used to reduce bias towards eutrophic lakes (Bennion et al., 1995), the transfer function does not perform well when there are dual gradients. An assumption of a transfer function is that any other variables that may affect the diatom assemblages have a negligible effect (Smol, 2008). This assumption cannot be met as diatoms are sensitive to numerous environmental variable, which is why they are so useful for environmental reconstruction. It is possible that the transfer function is reconstructing another variable instead of, or alongside TP (Battarbee et al., 2012). However, the transfer function overestimates all TP values, not just the lower TP which can occur due to edge effect. This means that, although the values are incorrect, the overall trend remains realistic, especially as the trend is similar to DCA1 and is likely to be the main driver of variation in diatom assemblages. There is also the issue of circular reasoning as the training sets are used to construct and test the transfer function (Smol, 2008), which must be taken into account. Despite the numerous limitations of using a transfer function to reconstruct the past environment of WLM, DI–TP results give the highest level of prediction compared to other phytoplankton TP estimates (Lotter et al., 1998). The mean bias of the TP (the difference between the inferred and measured TP during cross validation) was only 0.005 log10 µg/L, which is similar to other studies and lower than Bradshaw's (2002), which was 0.07 log10 µg/L. Development in training sets, further study into diatoms ecological optimas and acknowledgment of limitation allow diatom based transfer function to perform robustly (Smol, 2008).

Ecosystem Recovery

The crannog had numerous effects on the lake ecosystem and the differences discussed between WLM–1 and WLM–3 would suggest there has not been a full recovery. The phase diagram shows three separate spaces; pre, post and during the crannog phase (Figure 9). This again is an indication that the ecosystem has not returned to its original pre–crannog position.

Figure 10. Left: Phase diagram of White Loch of Myrton. Right: Phase space showing three key zones, modified from Batterbee et al., (2012) diagram

The phase diagram shows that the deterioration (Pathway D) and recovery trajectories (Pathway E) do not follow the same path. This is seen in many shallow lakes, and convoluted trajectories that lead to a different end point are common (Bennion et al., 2015). Nutrient reduction in shallow lakes is particularly complex, resulting in few showing a linear recovery trajectory. The concept that deterioration and recovery trajectories can vary in their paths can be seen in Figure 11 (Battarbee et al., 2012). Similarly to the phase diagram for White Loch of Myrton, the conceptual diagram has three phase spaces; 'a' showing the reference state; 'b' showing the maximum point of nutrient enrichment and 'c' showing the likely end point. Despite showing evidence of recovery, the system does not end at point 'a.' Battarbee et al.,'s (2012) conceptual diagram explains well the systems deterioration and recovery of the White Loch of Myrton.

A difference between the deterioration and conceptual trajectories is linked to hysteresis and alternate states (Beisner et al., 2003). The WLM reaching an alternate state may also explain the difference in the start and end phase spaces. Alternate states are common in eutrophic lakes and occur when there is an abrupt change in the ecosystem. Despite the increase in nutrients, a lake can be stable (Scheffer et al., 2001). There is an abrupt change between 64.5cm and 70.5cm, and it could be argued there is a slight folded curve which signifies an unstable equilibrium (Beisner et al., 2003), and 'b' is in fact an alternate state. However, the system does not meet Peterson's criteria of an alternate state as the system moves in to a different space when the pressure is removed (Capon et al., 2015). When the TP begins to decrease in WLM, a shift in space is apparent, thus disproving the idea that the White Loch of Myrton reached an alternate state.

There is an idea proposed by Hobbs et al., (2009) that directly compares human interference with ecosystem change (Figure 10). The historic ecosystem can change into hybrid or novel ecosystems, depending on the extent of the human disturbance. A hybrid ecosystem maintains some historic characteristics but species composition varies, whilst novel ecosystems have completely different functional properties. The dissimilarity results would suggest that a hybrid ecosystem has occurred post crannog. Although post crannog floristic change is significantly dissimilar, in comparison to the pre–crannog flora, there is evidence of historical characteristics, for example, the reappearance of Flagilaria spp.. There is also evidence of acid tolerant species decreasing post–crannog which is a historical characteristic and a sign of recovery in shallow lakes (Battarbee et al., 2014). Even in significantly recovered lakes, a complete reappearance in pre–enrichment flora is rarely seen (Bennion et al., 2015).

Figure 11.Three main system states proposed by Hobbs et al., (2009)

The concept of ecosystem regime shifts is still debated within the scientific community and many studies have faced analytical scrutiny. Some studies misuse terminology and conclude a system has reached an alternate state as opposed to the system actually being non–linear threshold dependant (Capon et al., 2015). Capon et al., (2015) concluded that the most common model for shallow lakes is an ecosystem that is threshold dependant with evidence of hysteresis. Whilst it has not been analytically proved, it is likely that the White Loch of Myrton faced a threshold dependant change during the crannog phase, rather than an alternate state. Despite the difference between the start and end point and the development of a hybrid ecosystem, there has been some recovery, and a novel ecosystem or alternate state has been avoided.

To understand the reason for a lack of full recovery, the post–crannog phase must be split into two subzones within WLM–3, which were highlighted in the cluster analysis (Figure 3). Between 70.5–28.5cm, there is a recovery period and at 20.5–4.5cm there is a second phase of poor ecological wellbeing and eutrophication. This secondary eutrophication is characterised by diversity decrease and increases in SCD and DCA1 scores. It also shows similarities with Gibson et al's (2003) study on Lower Lough Eire. The secondary eutrophication is simultaneous with a rise in TP, and is unlikely to be linked to the crannog phase and is therefore not discussed further. The inability to attain a full recovery to the same endpoint is often explained by internal loading of phosphorus and a continuation of high TP concentrations (Battarbee et al., 2012). Between 70.5–28.5cm the TP concentrations are low and therefore, TP cannot be used to explain the lack of full recovery in WLM as TP returns to a pre–crannog concentration of >106 µg/L. Whilst internal loading can delay recovery, the time span in which the core has been studied is greater than the time frame internal loading occurs.

Further Research

Further research on the impacts crannog usage had on the White Loch of Myrton's ecosystem could enhance understanding. This study examined the core at a low resolution, though over a long period of time. Low resolution within the recovery period was a limitation in this study and has been in others (Bennion et al.,2015), so a finer resolution would strengthen results. Low and high resolution proxy studies are often used alongside each other to create accurate reconstructions (Johnson, 2000). The low resolution results of this study have highlighted the key areas of interest; ecosystem transition points and recovery period. A high resolution study around these key areas would provide more accurate results.

Whilst some dates were available, an age–depth model would have shown changes in environmental conditions on a temporal scale (Blaauw and Christen, 2011). This would have provided important information such as; how long the crannog was used for and how long it took for the ecosystem to start recovering. Dating of the crannog would allow comparison with other Scottish crannogs, which are lacking radiometric dating, which would further understanding of Scottish crannogs.

The study could have been strengthened by using more proxies to provide a more sophisticated reconstruction. A good example of a crannog study that utilises all possible proxies is Ballywin Crannog (O'Brien et al., 2005), where palynology, chironomids, spores and macrofossils are investigated. Palynology may provide information on the human activity meaning the relationship between human activity and ecological change would be better understood.

Conclusion

The changes in the ecology of the White Loch of Myrton have been proven to be contemporaneous with the proposed crannog phase, with the crannog driving these changes. Numerous changes occurred that would suggest poor ecological wellbeing including changes in; species composition, species diversity and nutrients. The diatom assemblage was dominated by periphytic species within the crannog phase. Total phosphorus was shown to be the main environmental gradient which caused variation in the diatom assemblage. Total phosphorus increased at the depths the crannog was present, and is likely to be caused from an in–wash of waste products leaving the crannog.

The ecosystem does show a period of recovery, though a second period of modern nutrient enrichment does occur afterwards. The species composition appears to recover, with some species do not return though they are replaced with ecological equivalence species. Though the wellbeing of the ecosystem improves, a full recovery does not occur. The ecosystem that forms could be described as hybrid, with the square chord dissimilarity supporting this idea.

The sensitivity of diatoms to environmental conditions have provided useful, historic information on the ecosystem of the White Loch of Myrton. Diatoms highlight well changes in the ecological wellbeing of the lake. Whilst the DI–TP overestimates values, it provided key information on phosphorus trends. The study has further shown how powerful an indicator diatoms can be to environmental change. The study was strengthened by the use of multiproxy data, which supported diatom conclusions and explained some unusual results, including the lack of planktonic diatoms in the crannog phase. Overall the study has provided key information on the past environmental conditions on the disturbed White Loch of Myrton, in an area that is understudied and lacks understanding.

Acknowledgments

Firstly, I would like to thank my supervisor Prof. Pete Langdon for the support throughout my dissertation. Pete introduced me to palaeoecology during a second year semester and inspired me to undertake it as my dissertation topic. I would also like to thank Dr. Maarten van Hardenbroek and Thierry Fonville who have both shown a great deal of patience when introducing me to diatom identification, core analysis and transfer functions. Both Maarten and Thierry were always willing to spare some time to provide help, despite being incredibly busy.

References

Anderson, N. J., Rippey, B. and Gibson, C. E. (1993) A comparison of sedimentary and diatom–inferred phosphorus profiles: implications for defining pre–disturbance nutrient conditions. Hydrobiologia. (253) pp.357–366

Barber, J and Crone, B. A .(1993) Crannogs; a diminishing resource? A survey of the crannogs of South West Scotland and excavations at Buiston crannog. Antiquity. (67) pp.520–33

Battarbee, R. (1986). Diatom analysis. In: Berglund, B (ED.) Handbook of Holocene Palaeoecology and Palaeohydrology. Chichester. Wiley.

Battarbee, R. W., Anderson, N. J., Bennion. H. and Simpson, G.L. (2012) Combining limnological and palaeolimnological data to disentangle the effects of nutrient pollution and climate change on lake ecosystems: problems and potential. Freshwater Biology. (57) 10. pp.2091–2106

Battarbee, R. W., Anderson, N. J., Jeppesen, E. and Leavitt. P.R. (2005) Combining palaeolimnological and limnological approaches in assessing lake ecosystem response to nutrient reduction. Freshwater Biology. (50) 10. pp. 1772–1780

Battarbee, R. W., Simpson, G. L., Shilland, E. M., Flower, R. J., Kreiser, A., Yang, H. and Clarke, G. (2014). Recovery of UK lakes from acidification: an assessment using combined palaeoecological and contemporary diatom assemblage data. Ecological Indicators (37), pp.365–380

Batterbee, R. W., Jones, V. J., Flower, R. J., Camerons, N. G., Bennion, H., Caralho, L. and Juggins, S. (2001) Tracking Environmental Change Using Lake Sediments. Springer, Netherlands

Beisner, B. E., Haydon, D. T. and Cuddington, K. (2003) Alternative stable states in ecology. Frontiers in Ecology and the Environment. (1)7. pp.376–382

Bennion, H. (1995). Surface–sediment diatom assemblages in shallow, artificial, enriched ponds, and implications for reconstructing trophic status. Diatom Research. (10) pp.1–19.

Bennion, H., Simpson, G. L. and Goldsmith, B. J. (2015) Assessing degradation and recovery pathways in lakes impacted by eutrophication using the sediment record. Frontiers in Ecology and Evolution. (3)94. pp. 1–20

Bigler, C., von Gunten, L., Lotter, A. F., Hausmann, S., Blass, A., Ohlendorf, C. and Sturm, M. (2007) Quantifying human–induced eutrophication in Swiss mountain lakes since AD 1800 using diatoms. The Holocene. (17)8. pp.1141–1151

Birks, H. J .B. and Birks, H. H. (1980) Quaternary palaeoecology. The Blackburn Press. New Jersey.

Boulangeat, I., Lavergne, S., Van Es, J., Garruad, L. and Thuiller,W. (2012) Niche breadth, rarity and ecological characteristics within a regional flora spanning large environmental gradients. Journal of Biogeography. (39) p.204–214.

Bradbury, J. P. (1975) Diatom Stratigraphy and Human Settlement in Minnesota. GSA Special Papers. (171) pp. 1–74

Bradshaw, E. G., Anderson, J. N., Jensen, J. P. and Jeppesen, E. (2002) Phosphorus dynamics in Danish lakes and the implications for diatom ecology and palaeoecology. Freshwater Biology. (47) pp. 1963–1975.

Capon, S. J., Lynch, A. J., Bond, N., Chessman, B. C., Davis, J., Davidson, N., Finlayson, M., Gell, P. A., Hohnberg, D., Humphrey, C., Kingsford, R. T., Nielsen, D., Thomson, J. R., Ward, K. and Mac Nally, R., Regime shifts , thresholds and multiple stable states in freshwater ecosystems; a critical appraisal of the evidence. Science of the Total Environment. (534) pp.122–130

Cavers, G. (2006) Late Bronze and Iron Age lake settlement in Scotland and Ireland: the development of the 'crannog 'in the north and west. Oxford Journal of Archaeology (25)4. pp.389–412

Cavers. M.G. (2010) Black loch of Myrton, Monreith, Assessment Report. AOC Archaeology Group. Available: http://archaeologydataservice.ac.uk/archiveDS/archiveDownload?t=arch–414–1/dissemination/pdf/aocarcha1–82736_1.pdf Last Accessed: 20/08/2016

Crone. B.A., and Cavers. M.G. (2016) White Loch of Myrton, Wigtownshire Data Structure Report. AOC Archaeology Group. Available:

Davidson, T. A., Reid, M. A., Sayer, C. D. and Chilcott, S. (2013) Palaeolimnological records of shallow lake biodiversity change: exploring the merits of single versus multi–proxy approaches. Journal of Paleolimnology. (49) p.431–446.

Dean, W.E. (2006) Characterization of Organic Matter in Lake Sediments from Minnesota and Yellowstone National Park. USGS Open–File Report 2006.

Dobson, M. and Frid, C. (1998) Ecology of aquatic systems. Prentice Hall. Essex

EDDI. (1998). European Diatom Database. http://craticula.ncl.ac.uk/Eddi/jsp/index.jsp. Last accessed 30/01/2017

Fonville, T. (2015) Palaeoenvironmental Investigations of Crannogs in South–West Scotland and Co. Fermanagh, Northern Ireland. PhD Thesis. Southampton

Fredengren, C. (2002) Crannogs; a study of people's interaction with lakes, with particular reference to Lough Gara in the north–west of Ireland. Wordwell

Grimm, E. (2001). Tilia v1.6 computer program, Springfield: Illinois State Museum, Research and Collections Centre.

Groome. F. (1885) Ordnance Gazetteer of Scotland: A Survey of Scottish topography. Vol. VI. Edinburgh. Grange Publishing Works.

Hair, J. F., Black, W. C., Babin, B. J. and Anderson, R. E. (2010) Multivariate data analysis; a global perspective. 7 th Ed. Pearson. New Jersey

Hammer, Ø, Harper, D, and Ryan, P. (2001). PAST: Paleontonological statistics software package for education and data analysis. Palaeontologia Electronica

Henderson. J.C., Crone. B.A., and Cavers. M.G. (2003) A Condition Survey of Selected Crannogs In South West Scotland. Transactions of the Dumfriesshire and Galloway Natural History and Antiquarian Society. (77) p.79–102.

Hobbs, R. J., Higgs, E., and Harris, J. A. (2009). Novel ecosystems: implications for conservation and restoration. Trends in Ecology and Evolution. (24) pp.599–605.

Huber, K., Klee, R. and Schmidt, R. (2007) Distribution and morphological variability of Cyclotella–taxa in the late glacial of Längsee (Austria). Proceedings of the 1st Central European Diatom Meeting 2007.

Hurley, J. P., Armstrong, D. E., Kenoyes, G. J. and Bowse, C. J. (1985) Ground water as a silica source for diatom production in a precipitation–dominated lake. Science. (227) pp.1576–8.

Kondoh, M. (2001) Unifying the relationships of species richness to productivity and disturbance. Proceedings of the Royal Society of London. (268) pp.269–271

Krammer, K and Lange–Bertalot, H. (1999a ). Bacillariophyceae. 1. Teil: Naviculaceae. Süßwasserflora von Mitteleuropa, Band 2/1., Stuttgart, Gustav Fischer Verlag.

Krammer, K and Lange–Bertalot, H. (1999b). Bacillariophyceae. 2. Teil: Bacillariaceae, Epithemiaceae, Surirellaceae. Süßwasserflora von Mitteleuropa, Band 2/2., Stuttgart, Gustav Fischer Verlag.

Krammer, K and Lange–Bertalot, H. (2000). Bacillariophyceae. 3. Teil: Centrales, Fragilariaceae, Eunotiaceae. Süßwasserflora von Mitteleuropa, Band 2/3., Stuttgart, Gustav Fischer Verlag.

Krammer, K and Lange–Bertalot, H. (2004 ). Bacillariophyceae. 4. Teil: Achnanthaceae. Kritische Ergänzungen zu Achnanthes s.l., Navicula s.str., Gomphonema. Süßwasserflora von Mitteleuropa, Band 2/3., Stuttgart, Gustav Fischer Verlag.

Levinton, J. S. (1970) The paleoecological significance of opportunistic species. Lethaia. (3) pp.69–78

Little, J. L., Hall, R. I., Quinlan, R. and Smol, J. P (2000) Past trophic status and hypolimnetic anoxia during eutrophication and remediation of Gravenhurst Bay, Ontario: comparison of diatoms, chironomids, and historical records. Canadian Journal of Fisheries and Aquatic Sciences. (57) pp.333–341

Lobo, E. A., Wetzel, C. A., Ector, L., Katoh, K., Blanco, S. and Mayama, S. (2010) Response of epilithic diatom communities to environmental gradients in subtropical temperate Brazilian rivers Asociacion Iiberica de Limnologia. (29)2. pp.323–340

Lotter, A. F., Birks, J. B., Hofmann, W. and Marchetto, A. (1998) Modern diatom, cladocera, chironomid, and chrysophyte cyst assemblages as quantitative indicators for the reconstruction of past environmental conditions in the Alps. II. Nutrient. Journal of Paleolimnology. (19) pp.443–463

Lowe, J. J. and Walker, M. J. C. (2015) Reconstructing Quaternary environments. 3rd edition. Pearson Education Limited. Essex.

Lowe, R. E., du Buf, J. M. H., Bayer, M. M. and Mann, D. G. (2004) Diatom classification in ecological applications. Pattern Recognition. (37) p.1283–1285.

Lowe, R.L. (1996) Periphyton patterns in lakes. In Stevenson,R.J., Bothwell, M.L., and Lowe, R.L. Algal Ecology: Freshwater Benthic Ecosystems. Academic Press. San Diego.

Magurran, A. (1988) Ecological diversity and its measurements. Chapman and Hall. London

Matton, R. (2016) An investigation into the effects of medieval crannog construction on Lough Yoan, Northern Ireland using diatom analysis. BSc Thesis. Southampton

McCormick, P.V. and Stevenson, R.J. (1998) Periphyton as a tool for ecological assessment and management in the Florida everglades. Journal of Phycology. (34)5. p.726–733.

Moss, B. (2010) Ecology of Freshwaters: A View for the Twenty–First Century. 4 th Edition. Wiley Blackwell. Chichester.

O'Brien, C., Selby, K., Ruiz, Z., Brown, A., Dinnin, D., Caseldineh, C., Langdon, P. and Stuijts, I. (2005) A sediment–based multiproxy palaeoecological approach to the environmental archaeology of lake dwellings (crannogs), central Ireland. The Holocene. (15) 5. pp.707–719

Palmer, M. W. (1993) Things in Even Better Order: The Advantages of Canonical Correspondence Analysis. Ecology. (74)8. pp. 2215–2230

Peet, R. K. (1974) The measurement of species diversity. Annual Review of Ecology and Systematics. (5) pp. 285–307

Peinerud, E. (1997) Diatom Silica in Lake Sediments and Suspended Matter. Lulea University of Technology, Licenticate Thesis. (34) pp.1–50

R: A Language and Environment for Statistical Computing (2015). R Core Team. R Foundation for Statistical Computing. Vienna, Austria.

Randsalu–Wendrup, L., Conley, D. J., Carstensen, J. and Fritz, S.C. (2016) Paleolimnological records of regime shifts in lakes in response to climate change and anthropogenic activities. Journal of Paleolimnology. (56) p.1–14

Renberg, I. (1990) A procedure for preparing large sets of diatom slides from sediment cores. Journal of Paleolimnology. (4) pp.87–90.

Ryves, D. B., Anderson, N. J., Flower, R. J. and Rippey. B (2013) Diatom taphonomy and silica cycling in two freshwater lakes and their implications for inferring past lake productivity. Journal of Paleolimnology. (49) 3. pp.411–430.

Scheffer, M. (2004) Ecology of Shallow Lakes. Kluwer Academic Publishers. AA Dordrecht, The Netherlands.

Scheffer, M. and Carpenter, S. R. (2003) Catastrophic regime shifts in ecosystems: linking theory to observation. Trends in Ecology and Evolution (18)12. pp.648–656

Scheffer, M., Carpenter, S., Foley, J., Folke, C. and Walker, B. (2001). Catastrophic shifts in ecosystems. Nature. (413) pp.591–596

Selby, K. A. and Brown, A. G. (2007) Holocene development and anthropogenic disturbance of a shallow lake system in Central Ireland recorded by diatoms. Journal of Paleolimnology. (38) p.419–440.

Smol, J. P (2008) Pollution of Lakes and Rivers; A Palaeoenvironmental Perspective. 2 nd Edition. Wiley Blackwell.

Smol. J. P and Stoermer. E. F. (2010) The Diatoms; a primer in The Diatoms: Applications for the Environmental and Earth Sciences. 2 nd Edition. Cambridge University Press. Cambridge.

Søndergaard, M., Jensen, J. P. and Jeppesen, E. (2003) Role of sediment and internal loading of phosphorus in shallow lakes. Hydrobiologia (506–509) pp.135–145

Toporowska, M., Pawlik–Schow RonSka, B. and Kornijow, R. (2010) Winter versus summer blooming of phytoplankton in a shallow lake: effect of hypertrophic conditions. Polish Journal of Ecology. (58)1. pp.3–12

University of Manchester, No Date. Research Projects. Available: http://www.arts.manchester.ac.uk/archaeology/research/dunragit/ Last Accessed: 31/01/2017

Vadeboncoeur, Y., Jeppesen, E., Zanden, M. J. V., Schierup, H., Christoffersen, K. and Lodge, D. M. (2003) From Greenland to green lakes: Cultural eutrophication and the loss of benthic pathways in lakes. American Society of Limnology and Oceanography. (48)4. p.1408–1418.

Witkowski, A. and Pempkowiak. (1995) Reconstructing the development of human impact from diatoms and 210PB sediment dating (The Gulf of Gdarisk– Southern Baltic Seas). Geographia Polonica (65) pp.63–78