Response of chironomids to Neolithic land-use change in north-west Ireland

Coring and sub-sampling

The two parallel 13-m-long cores were extracted using a Usinger piston corer, at a water depth of 6 m close to the centre of Templevanny Lough in 2008 (Stolze et al., 2013b). The sediment interval between 999.5 and 839.5 cm was contiguously sampled at 1-cm resolution for chironomid, geochemical, macroscopic charcoal and pollen analyses.

Radiocarbon dating

Age control for Templevanny Lough is based on 16 ¹⁴C accelerator mass spectrometry (AMS) determinations from Stolze et al. (2013b). An additional six dates were obtained for this study using terrestrial plant material retrieved from the sediments (Stolze et al., in preparation). The sediment samples were treated with 10% HCl and rinsed with deionized water through a 125-µm sieve. The plant material was picked under a stereo microscope. Radiocarbon dating was performed at the ¹⁴CHRONO Centre, Queen’s University Belfast. A new depositional model was calculated for the 160-cm-long sediment sequence with the program OxCal 4.2 (Bronk Ramsey, 1995, 2001, 2009), applying the IntCal13 curve (Reimer et al., 2013). Calculation of the Poisson depositional model was performed at 0.5-cm increments and a number of events per unit length of 10 cm⁻¹. Boundaries were set at 975.5 and 897.5 cm. Outlier analysis was performed and identified 6 of the 22 samples as not reliable. Ages obtained by Bayesian age–depth modelling are expressed as the mean of the posterior probability distributions and rounded to the nearest 10 years. The 95.4% highest probability density intervals are used as a measure of uncertainty of this age estimate, which range from 25 to 85 years, with a median value of 40 years.

Chironomid analysis

Chironomid analysis followed standard procedures outlined by Walker (2001). Between 0.75 and 4.5 cm³ of sediment was deflocculated in a 10% KOH solution and heated at 30°C for 30 min. The sediment slurry was sieved through a 90-µm mesh, and the sieve residue was poured into a Bogorov plankton counting tray to be sorted at 10–40× magnification using a Motic^® SMZ series dissection microscope. A minimum of 50 head capsules per sample were removed with forceps and permanently mounted on slides with Entellan^® for identification. Chironomid identifications were made using a Motic B3 professional series compound microscope at 100–400× magnification. Taxa were identified to genus, sub-genus or species level following Brooks et al. (2007), Larocque-Tobler (2014), Rieradevall and Brooks (2001) and Wiederholm (1983). Chironomid ecology and distribution information are based on Brooks et al. (2007), Moller Pillot (2009, 2013) and Vallenduuk and Moller Pillot (2007).

For presentation purposes, chironomid types associated with rivers and streams were grouped together in the chironomid stratigraphy. Total stream taxa include Corynoneura lobata-type, Epoicocladius, Euryhapsis, Heleniella, Krenosmittia, Paracricotopus, Parametriocnemus/Paraphaenocladius, Paratrissocladius, Stilocladius, Rheocricotopus undifferentiated, Rheocricotopus chalybeatus-type, Rheocricotopus effusus-type, Rheocricotopus fuscipes-type and Thienemanniella clavicornis-type.

Geochemical analyses (δ¹³C, δ¹⁵N and C:N)

All contiguous samples were analysed for stable isotope ratios of nitrogen and organic carbon (δ¹⁵N and δ¹³C) and the elemental concentrations of C and N. Sediment samples of 1 cm³ were dried, pulverized with a mortar and pestle and passed through a 200-µm sieve. Between 2 and 5 mg of homogenized material was weighed into silver capsules and HCl-fumigated following Harris et al. (2001) and measured for δ¹³C values. A separate analysis was conducted to measure δ¹⁵N values of a non-acidified portion of the samples. Stable isotope ratios of C and N were determined using a Thermo Fisher Delta V Advantage elemental analysis–isotope ratio mass spectrometry (EA-IRMS) at the Isotope Biogeochemistry Laboratory, University of Hawai’i at Mānoa. C:N was calculated from organic carbon and total nitrogen (TN) elemental concentrations. Isotope ratios are reported as ‰ relative to the Vienna PeeDee Belemnite (VPDB) standard and atmospheric nitrogen for δ¹³C and δ¹⁵N, respectively. Accuracy, that is, the offset between measurements and standards, for both δ¹³C and δ¹⁵N, was ±0.3‰.

Macroscopic charcoal analysis

For all 160 samples, macroscopic charcoal particles (⩾90 µm) were counted in concert with the chironomid extraction, and charcoal accumulation rates (particles cm⁻² yr⁻¹) were determined.

Loss-on-ignition and pollen analyses

To provide a complete pollen and geochemical record for the Neolithic period from Templevanny Lough, the previously published loss-on-ignition (LOI) and pollen stratigraphies (Stolze et al., 2013b) were each complemented by the analysis of 32 additional sediment samples taken between 870.5 and 839.5 cm. Detailed preparation protocols for both analyses are given in Stolze et al. (2013b). The organic (LOI₅₅₀) and inorganic carbon (LOI₉₅₀) contents determined at 550°C and 950°C (Heiri et al., 2001) as well as the non-combustible fraction (NCF) are presented here. In the present contribution, a percentage curve summarizing pollen types largely indicative of grassland communities (cf. Behre, 1981; Brun, 2011) and the arboreal pollen curve are displayed. These pastoral pollen indicators (PPIs) include Cirsium, Tubuliflorae, Lactucae, Plantago lanceolata-type, Poaceae: wild grass group, Rumex acetosa-type, Plantago major–media-type, Ranunculus-type and Trifolium-type. In addition, the Plantago lanceolata-type curve is shown. Due to the nearly consistent occurrence of Hordeum-type pollen in the pollen stratigraphy, possibly reflecting the presence of Glyceria on the lake fringe rather than the cultivation of Hordeum (cf. Stolze et al., 2012), this pollen type is not considered as an anthropogenic indicator here. Calculation of the pollen percentages is based on the terrestrial pollen and spore sum.

Statistical analyses

Ordination analyses were performed using Canoco version 4.54 (Ter Braak and Šmilauer, 2006) on square root transformed chironomid percentage data for all common taxa, that is, taxa present in at least two samples with a relative abundance of ⩾2% in at least one sample (Quinlan and Smol, 2001). Both principal component analysis (PCA) and detrended correspondence analysis (DCA) were carried out. PCA results were used as this analysis gave a better overall explanation of the variance in the data. Redundancy analysis (RDA), a constrained linear ordination technique, was performed to identify which of the environmental variables could explain a statistically significant (p < 0.05) amount of the variance in the chironomid data set. The statistical significance of each variable was assessed using Monte Carlo permutation tests (499 unrestricted permutations). A series of partial RDAs was completed for six environmental variables (PPI, Plantago lanceolata-type, macroscopic charcoal, C:N, δ¹⁵N and δ¹³C) to determine which of the variables retained high explanatory importance for chironomid distribution.

The chironomid percentage diagram was constructed using C2 version 1.7.6 (Juggins, 2014), and zonation was performed on the percentage data of all taxa using Zone version 1.2 (Juggins, 1992). Zonation was based on optimal sum-of-squares partitioning, and statistical significance of zones was determined using BSTICK (cf. Bennett, 1996). All other graphs and linear regressions were completed using SigmaPlot version 12.0.

Chironomid stratigraphy and zonation

In total, four statistically significant zones were determined. The subdivision of Zone 4 is not statistically significant but was included here as the sub-zonation shows a noteworthy qualitative change within the chironomid community. Chironomid head capsules of 50–82.5 (mean = 56) were identified per sample. A total of 126 distinct taxa were identified in the Templevanny Lough core, with 19–40 different taxa (mean = 29) per sample.

Stable carbon and nitrogen isotopes and LOI

Values of lake sediment δ¹⁵N range from 2.3–5.4‰ with a mean of 3.9‰ and TN shows a range of 0.9–1.7% (mean = 1.3%; Figure 4). The values for δ¹³C vary from −30.5‰ to −32.6‰, with a mean of −31.6‰. C:N ratios of organic matter show a range of 11.3–14.6, with a mean value of 13.3. LOI₅₅₀ values range from 19.8–43.7% (mean = 32.7%), and NCF values vary from 44.9–61.7% (mean = 51.5%). The zones established for the chironomid data were also applied to the other proxies to allow for an easier comparison of the various data sets.

Pollen and macroscopic charcoal

Macroscopic charcoal accumulation rates range from 0.1 to 8 particles cm⁻² yr⁻¹, with a mean value of 1.9 particles cm⁻² yr⁻¹ (Figure 4).

RDA

RDA produced eigenvalues for Axis 1 of λ = 0.093 and for Axis 2 of λ = 0.040. Analysis identified that the six chosen environmental variables (PPI, P. lanceolata-type, macroscopic charcoal, δ¹⁵N, δ¹³C and C:N) have statistically significant (p < 0.01) relationships with the chironomid composition (Figure 5; Table 1).

OPEN IN VIEWER

Table 1.

Partial redundancy analyses (RDAs) for environmental variables by themselves and with the effects of other environmental variables partialled out.

Variable	Covariables	λ1	λ1/λ2	p	% variance	PCA Axis 1 as predictor (r2)
PPI	None	0.084	1.292	0.002	8.4	0.689*
	Charcoal, δ15N, δ13C, C:N	0.021	0.447	0.002	2.5
Plantago lanceolata-type	None	0.061	0.801	0.002	6.1	0.444*
	Charcoal, δ15N, δ13C, C:N	0.014	0.298	0.002	1.7
δ15N	None	0.041	0.373	0.002	4.1	0.054*
	Charcoal, Plantago lanceolata-type, PPI, δ13C, C:N	0.022	0.468	0.002	2.6
δ13C	None	0.043	0.518	0.002	4.3	0.312*
	Charcoal, Plantago lanceolata-type, PPI, δ15N, C:N	0.013	0.277	0.002	1.5
C:N	None	0.060	0.870	0.002	6.0	0.470*
	Charcoal, Plantago lanceolata-type, PPI, δ15N, δ13C	0.015	0.319	0.002	1.8
Macroscopic charcoal	None	0.062	0.775	0.002	6.2	0.434*
	Plantago lanceolata-type, PPI, δ15N, δ13C, C:N	0.011	0.234	0.002	1.3

PPI: pastoral pollen indicator; PCA: principal component analysis.

This table also includes for each variable a linear regression correlation co-efficient with chironomid PCA Axis 1.

*

p value of <0.01.

Partial RDAs show that PPI exhibited the strongest correlation with the chironomid community, explaining 8.4% of the variance on its own and 2.5% with other variables partialled out (Table 1). A linear regression further highlights a strong relationship between PCA Axis 1 and the PPI data (r² = 0.689, p < 0.01; Table 1). Macroscopic charcoal (6.2% variance), P. lanceolata-type (6.1% variance) and C:N (6.0% variance) also exhibited strong relationships with the chironomid assemblage data (Table 1).

The RDA bi-plot highlights that PPI and P. lanceolata-type are likely the main variables behind the chironomid community change in Zone 2 (Figure 5). C:N and δ¹³C show a relationship with the chironomid assemblage of Zone 4, with higher values in the environmental variables corresponding with Zone 4 samples in the bi-plot. RDA analysis also shows that the chironomid community in Zone 1 correspond with the low δ¹⁵N values.

Zone 1 – Early anthropogenic disturbance and onset of Neolithic farming

High levels of macroscopic charcoal recorded from c. 4110–4000 BC, indicate local natural or anthropogenic fire activity (Whitlock and Larsen, 2001), pre-dating widespread forest clearance in the region (cf. O’Connell et al., 2014; Stolze et al., 2013b). Microscopic and macroscopic charcoal were also present around this time at Cooney Lough (O’Connell et al., 2014) and Lough Dargan (Taylor et al., 2013), both located within a c. 20 km radius of the study site. The timing of this peak in macroscopic charcoal also coincides with the construction dates for the early causewayed enclosure at Magheraboy, County Sligo (Danaher, 2007). Together with the presence of Pteridium aquilinum spores (Stolze et al., 2013b) and occasional P. lanceolata-type pollen, the record points to human-related disturbances. Chironomids do not exhibit a noticeable response to this early activity, as it was likely not at a scale that would influence lake trophic status. However, high C:N ratios and increasing δ¹³C values infer greater terrestrial inputs during this time.

An initial decrease in the forest cover occurs between c. 3970 and 3850 BC and coincides with the mid-Holocene elm decline (Stolze et al., 2013b). The gradual increase in δ¹⁵N and NCF and decrease in C:N suggest nutrient influx and enhanced productivity of Templevanny Lough as a result of soil destabilization due to loss of canopy (cf. Cohen, 2003). The chironomid community shows a subdued response to this vegetational change. Cladopelma, which is not particularly tolerant of high nutrient conditions (Brooks et al., 2007; Moller Pillot, 2009), disappears by c. 3930 BC, only to return sporadically in low numbers for the remainder of the Neolithic period.

Increased fire activity at c. 3860 BC, as evident in the macroscopic charcoal record, precedes the accelerated deforestation at c. 3850 BC. The pronounced increase in PPI, in particular P. lanceolata-type pollen, from c. 3850–3820 BC indicates the onset of Neolithic pastoral farming activity in the catchment area of Templevanny Lough. The chironomid community composition shows a markedly stronger response to pastoral farming activity than to the previous loss in forest cover linked to the elm decline. However, the chironomids seem to experience a lagged response to the anthropogenic activity. The lagged effect occurs throughout the sampling range by 1–2 samples (~10–20 years). For example, the resulting impact of an increase in PPI from 3850–3820 BC is seen two samples later by an increase in Tanytarsus mendax-type from 5% to 17% and T. lugens-type from 0% to 6% from c. 3810–3800 BC. Tanytarsus mendax-type, a warm stenotherm has previously been associated with pastoral agriculture in western Ireland (Potito et al., 2014; Taylor et al., 2013). Tanytarsus lugens-type, however, is usually associated with oligotrophic lakes (Brooks et al., 2007; Saether, 1979) but appears to be responding to the increased nutrient input at Templevanny Lough. This taxon also shows an association with warm, productive lakes in western Ireland surface sediments (Potito et al., 2014). Wilson and Gajewski (2004) noted that T. lugens-type was common in shallow, productive lakes in Northern British Columbia and suggest the possibility that T. lugens-type has a more widespread ecological distribution than previously thought. Cool summer conditions in Ireland (~16°C July temperature average; Met Éireann, 2016) may result in T. lugens-type exhibiting a similar ecological response to nutrient enrichment, as T. lugens-type in colder regions. However, this taxon has been associated with cold late glacial conditions in a small lake in County Donegal, Ireland (Watson et al., 2010), suggesting that the full ecological distribution of T. lugens-type in Ireland may merit further investigation. Lag effects between land-use change and lake eutrophication could be due to within-lake nutrient cycling or delayed nutrient inwash (Genkai-Kato and Carpenter, 2005). Response times can be highly variable and are affected by catchment morphology and soil type, lake morphometry, temperature and dominance of macrophytes, among other factors (Genkai-Kato and Carpenter, 2005). In the short term, accumulation of nitrogen and phosphorus in the soils from manure is an important driver of nutrient input. Nutrient-rich soils are washed into lakes, where some of the phosphorus dissolves and stimulates growth of phytoplankton and aquatic plants (Carpenter, 2005). Within-lake recycling of nutrients from sediments and consumers can then act to enhance eutrophication through time (Carpenter, 2005; Carpenter et al., 1998). Templevanny Lough geochemistry does not show a lag with land-use change, so within-lake nutrient cycling, rather than delayed nutrient inwash, is likely responsible for the lagged chironomid response.

Chironomid larvae do not exploit nutrients directly but instead indirectly respond to changes in the availability of food (mainly diatoms, organic detritus and macro-invertebrates) and benthic oxygen conditions (Brooks et al., 2001). Chironomids can exhibit a gradual or threshold response to nutrient inputs, and response time to cultural eutrophication can often lag land-use change (Langdon et al., 2006). Lotter and Birks (2003) note a lag of several centuries in the chironomid community in response to changes within an aquatic ecosystem of a small lake, Sägistalsee in the Swiss Alps. A lagged chironomid response was not visible in the Lough Dargan record, presumably as a result of the low sampling resolution of 35 to 270 years (mean = 70 years; Taylor et al., 2013). Interestingly, lagging Templevanny Lough PPI and P. lanceolata-type pollen by two samples (~20 years) results in marginally more powerful partial RDAs (8.5% explained variance for PPI; 7.5% explained variance for P. lanceolata-type) and stronger relationships with chironomid PCA Axis 1 (r² = 0.713 for PPI; r² = 0.586 for P. lanceolata-type) than the non-lagged models used in this paper.

Zone 2 – Early Neolithic farming

The most intensive period of Neolithic farming activity at Templevanny Lough coincides with high levels of archaeological activity found in Ireland between c. 3750 and 3500 BC, including the construction of rectangular houses between 3715 and 3625 BC (Whitehouse et al., 2014). The causewayed enclosure at Magheraboy and the Carrowmore passage tomb complex provide evidence of human activity in the Sligo area during the early Neolithic (Bergh, 1995; Danaher, 2007; Hensey and Bergh, 2013). Neolithic farming at Templevanny Lough precedes dates from Carrowkeel-Keshcorran cairns by several centuries, although the Carrowkeel-Keshcorran dates may be compromised by insecure stratigraphic contexts (Hensey, 2015).

High levels of PPI and P. lanceolata-type indicate sustained open pasture between c. 3730 and 3690 BC (Figure 4). High δ¹⁵N values (4.1–4.6 ‰), together with the decrease in δ¹³C and low C:N suggest considerable nutrient enrichment in the lake system. High δ¹⁵N values in concert with decreasing levels of TN and δ¹³C are often indicative of pastoral farming around a lakeside and the inwash of animal waste into a lake (Botrel et al., 2014; Taylor et al., 2013; Woodward et al., 2012). A similar geochemical pattern of increased δ¹⁵N, low δ¹³C concomitant with high farming activity was recorded at Lough Dargan, supports the assumption of animal husbandry with substantial levels of animal waste reaching the lake during the Neolithic (Taylor et al., 2013). Furthermore, δ¹³C in modern lake sediment samples from agricultural catchments in western Ireland were more negative in value than δ¹³C in samples from non-impacted sites (Woodward et al., 2012). Averaged δ¹⁵N (4.4‰), δ¹³C (−31.8‰) and C:N (12.4) values place the early Neolithic at Templevanny Lough among lakes with >80% agricultural cover (mostly pastoral) within a 50-lake training set of modern lakes in western Ireland (Woodward et al., 2012).

The chironomid community in Zone 2 shows eutrophic taxa such as Chironomus anthracinus-type, Tanytarsus undifferentiated and Tanytarsus lugens-type dominating the stratigraphy (Figure 2). T. lugens-type is associated with modern agricultural activity in a western Ireland training set (Potito et al., 2014). T. lugens-type appears to have taken advantage of the new ecological opportunities following the onset of early Neolithic farming and rapidly dominates the chironomid stratigraphy. A decrease in T. lugens-type towards the end to this phase suggests that T. lugens-type may have been at a competitive disadvantage, eventually becoming replaced by other eutrophic taxa, mainly Chironomus anthracinus-type, an opportunistic eutrophic taxon linked to agriculture in western Ireland (Potito et al., 2014). The apparent decrease in Corynoneura lobata-type and Corynoneura type A appears to follow well with the decline in total stream taxa during Zone 2. Corynoneura as a genus can be found in a variety of different water bodies such as lakes, pools, streams, springs and marshes, with Corynoneura lobata-type in particular, being an inhabitant of running water (Moller Pillot, 2013).

Zone 3 – Decline in farming activity

Farming activity in the Templevanny Lough catchment wanes during the middle Neolithic, with periods of small-scale activity. Lower δ¹⁵N values in Zone 3 correspond with reduced farming activity as reflected in the PPI (Figure 4). Increasing C:N shows decreased lake productivity through time. This is in keeping with a lull in human activity, with settlement evidence appearing more ephemeral in Ireland between c. 3400 and 3000 BC (Whitehouse et al., 2014). Although farming was declining at Templevanny Lough, human impact on the lake system persisted, with Chironomus anthracinus-type and Tanytarsus undifferentiated remaining dominant during this time. This is expected, as lakes are known to take decades to centuries to recover from cultural eutrophication (Carpenter, 2005).

The PPI, P. lanceolata-type pollen, δ¹⁵N and macroscopic charcoal data indicate two main intervals of human activity in the area during the middle Neolithic at c. 3600–3560 BC and c. 3390–3300 BC. Small fluctuations in the abundances of Tanytarsus undifferentiated and Chironomus anthracinus-type suggest increased nutrient input as a result of this small-scale farming activity. This evidence shows that although farming activity is reduced, humans remain an active part of the landscape during this time.

Based on the proxy data, Templevanny Lough remains productive during the middle Neolithic likely due to sustained farming and/or prolonged lake recovery from earlier farming activity. The main chironomid taxa present during this time, including Chironomus anthracinus-type, Micropsectra insignilobus-type, Tanytarsus mendax-type, Tanytarsus undifferentiated, Paratanytarsus, Corynoneura undifferentiated, Corynoneura type A, Limnophyes/Paralimnophyes and Phaenopsectra flavipes-type, indicate a macrophyte-rich, eutrophic lake. Floating leaf macrophytes occurred in the lake throughout this period (Stolze et al., 2013b).

Micropsectra insignilobus-type appears anomalously in this stratigraphy as it is a cold stenotherm associated with oligotrophic lakes (Brooks et al., 2007). However, M. insignilobus-type sub-fossils are indistinguishable from M. atrofasciata-type (Brooks et al., 2007), which occurs in eutrophic rivers and increases below sewage discharges (Brooks et al., 2007; Wilson, 1989). Ruiz et al. (2006) also noted increases in M. insignilobus-type with nutrient enrichment in an investigation of a palaeochannel in the East Midlands, UK, and altered the identification of the M. insignilobus-type to M. atrofasciata-type based on its ecology.

Sub-zone 4a – Low farming activity during the late Neolithic

The palynological data and δ¹⁵N indicate low levels of farming activity and woodland recovery during most of the late Neolithic, with the exception of an increase in activity around c. 2980–2910 BC (Figure 4). A significant decrease in the abundance of the eutrophic taxon Chironomus anthracinus-type clearly reflects this downturn. Averaged δ¹⁵N (3.7‰), δ¹³C (−31.4‰) and C:N (14.2) values place this sub-zone, excluding the increase in values between c. 2980 and 2910 BC, among lakes with 20–80% agricultural cover within the 50-lake training set of modern lakes in western Ireland (Woodward et al., 2012). The chironomid community is largely dominated by taxa associated with macrophytes and/or terrestrial/semi-terrestrial habitat, including Microtendipes pedellus-type, Limnophyes/Paralimnophyes, Corynoneura undifferentiated, Corynoneura type A and Symposiocladius (cf. Stolze et al., 2013b).

The δ¹⁵N, P. lanceolata-type pollen and PPI data indicate a short-lived period of small-scale pastoral farming for 70 years between c. 2980 and 2910 BC, peaking at c. 2940 BC. Despite the ephemerality of this interval, the impact on the lake system is evident particularly in the geochemical data with fluctuations in δ¹⁵N, δ¹³C, C:N and LOI₅₅₀. C:N ratios infer increased lake productivity during this time, δ¹⁵N is elevated and LOI₅₅₀ decreases by 7%, possibly indicating inwash of allochthonous material (cf. Stolze et al., 2013b). The chironomid response to this period of small-scale farming activity is more subtle, with slight increases in Chironomus anthracinus-type, Tanytarsus undifferentiated and Micropsectra insignilobus-type during this time. This period of farming coincides with an interval of possible activity at the Carrowkeel-Keshcorran complex from approximately c. 3200–2900 BC, as evidenced by human bone remains and dated charcoal (Hensey et al., 2013). This period (3200–2900 BC) also represents a critical time in the passage tomb construction and use in the eastern part of Ireland (Hensey et al., 2013).

Sub-zone 4b – Return of farming at the end of the Neolithic period

Increases in δ¹⁵N, PPI and macroscopic charcoal signify a return of small-scale farming activity around Templevanny Lough. Sub-zone 4b marks the possible build up to the early Bronze Age farming in the Carrowkeel-Keshcorran area (cf. Hensey et al., 2013). The C:N ratio decreases at Templevanny Lough, suggesting an increase in autochthonous productivity from c. 2600 BC. A decrease in LOI₅₅₀ and increase in NCF values suggest influx of allochthonous material, possibly leading to nutrient enrichment of the lake system.

The δ¹³C values fluctuate substantially during the later Neolithic. The elevated values between c. 2660 and 2610 BC could be indicative of a possible inwash event, changes in lake level or an increase in macrophyte productivity. Increased C:N ratios during this time provide evidence of terrestrial material being deposited into the lake, which is supported by high NCF. δ¹⁵N, TN and the presence of P. lanceolata-type pollen and macroscopic charcoal suggest that this increase in δ¹³C is not directly land-use related. Chironomid taxa associated with terrestrial material, including Limnophyes/Paralimnophyes, Smittia and Symposiocladius, are present during this time. Furthermore, the increase in total stream taxa and δ¹³C values likely indicates increased inflow from the surrounding streams.

The chironomid community indicates a eutrophic, macrophyte-rich lake with the presence of Chironomus anthracinus-type, Chironomus plumosus-type, Tanytarsus undifferentiated, Paratanytarsus, Endochironomus albipennis-type, Glyptotendipes pallens-type and Cricotopus intersectus-type in various lagged responses to the farming activity. The emergence of Chironomus plumosus-type and an increase in δ¹⁵N from c. 2530 BC coincide in an increase in PPI at c. 2550 BC, signifying substantial nutrient input to the lake system. Chironomus plumosus-type is a eutrophic taxon that has been used as an indicator of hypertrophic conditions in Danish lakes (Brodersen and Lindegaard, 1999).

Palaeolimnological impacts of Neolithic farming

The timing and magnitude of changes in the chironomid stratigraphy show that land-use, potentially influenced by climate change (Stolze et al., 2013b), is the main control on the chironomid community at Templevanny Lough during the Neolithic. Although chironomids have been mostly used as a palaeotemperature indicator in the past, as temperature tends to be the main driver behind chironomid community compositional change (Brooks and Birks, 2001), more recent studies have shown that even moderate human impacts are sufficient enough to over-ride the climate signal in the chironomid record (Brodersen and Quinlan, 2006; McKeown and Potito, 2016), which seems to be the case in Templevanny Lough.

RDA and partial RDAs show a strong statistical relationship between PPI and P. lanceolata-type pollen with the changes in the chironomid composition, with the relationship especially apparent for Zone 2 (Figure 5). The strong correlation between PCA Axis 1 and PPI further highlights that pastoral land-use was a dominant influence on chironomid community compositional change throughout the record. Additionally, the chironomid community as reflected in the PCA Axis and the main chironomid taxa (e.g. T. lugens-type) show the same structural coherence as the PPI data in a lagged response (Figure 4). The chironomid community composition thus exhibited a predictable and proportional lagged response of c. 10–20 years after the farming activity occurred. If the chironomid community was directly driven by climatic forcing, the composition would change before or at the same time as any increase or decrease in human activity occurred in the catchment area because farming around Templevanny Lough during the Irish Neolithic was shown to follow climatic variation closely (Stolze et al., 2013b).

Organic residue analysis from pottery shows cattle dairying to be especially prominent in this area during the Neolithic (cf. Smyth and Evershed, 2015), so it is likely that lake trophic status is tied to pastoral activity of cows during this time. As the palaeolimnological proxies show a scaled response, rather than a simple threshold response, to Neolithic land-use change (Langdon et al., 2006), they can be used as reliable proxies for prehistoric land-use intensity. However, the chironomid response was systematically lagged throughout the record, likely due to within-lake nutrient dynamics (cf. Carpenter, 2005). Therefore, palynological evidence should be used to represent and date the onset and timing of Neolithic farming, whereas palaeolimnological evidence can show the intensity and environmental impacts of these activities.

Early Neolithic farming resulted in environmental impacts and freshwater eutrophication that would put Templevanny Lough on a par with lakes from modern pastoral catchments in the region (Woodward et al., 2012). The considerable impacts could be due to scale of activity (i.e. sizeable cattle herds) or location of grazing activity (i.e. adjacent to the lakeside). Either way, evidence from Templevanny Lough and Lough Dargan (Taylor et al., 2013), both in fertile lowland areas, shows that prehistoric pastoral farming practices regularly result in lake eutrophication across the region.