Water quality data for site across England were obtained for river sites from the UK Environment Agency's Water Quality Archive (Environment Agency, 2020). The dataset consisted of river water quality measurements collected at irregular intervals from 2000–2020. We focused on nitrate-N concentrations as the dominant form of dissolved N (NO3-–N). Initial screening of the data identified 21 049 sampling sites with NO3-–N data. Daily mean discharge records were obtained from the National River Flow Archive (NRFA) which contains daily discharge measurements for over 1500 UK gauging stations.

We applied the following criteria to filter and select nitrate data for our study: (1) the data needed to cover at least 80 % of the months between 2000 and 2019, (2) the time series could not include gaps longer than 3 consecutive months, (3) each water quality site must be within 1 km of a flow gauge with daily discharge (Q) data and located on the same river, and (4) the flow gauge must have at least 90 % of the discharge data available over the 20 year period. Application of the criteria above resulted in 66 catchments (Fig. S1 in the Supplement) with nitrate and corresponding discharge time series data. In total, there were 18 947 nitrate concentration observations and 482 130 flow measurements across these 66 sites and the 20 years of study. Short missing segments of daily discharges were filled using simple linear interpolation to produce the continuous daily series, but these gaps were rare and short relative to the full record. As a sensitivity check, we compared the months of annual maximum and minimum monthly mean discharge before and after discharge gap filling; the identified discharge-extreme month differed in only 13 of 1254 site-years (1.0 %), indicating limited influence on the synchrony classification. All data analyses were based on the water year (from 1 October–30 September of the subsequent year).

The selected catchments spanned a wide range of hydrological, topographical, land cover, lithology and soil characteristics and other descriptors (Table 1). A complete description of the hydrological variables now appears in Sect. S1 in the Supplement. The Standardized Precipitation Index (SPI) was obtained for 56 out of 66 studied catchments from the UK Centre for Ecology and Hydrology (UKCEH), via the UK Water Resources Portal (UK Centre for Ecology and Hydrology, 2024). The standard precipitation index (SPI) expresses precipitation anomalies in units of standard deviation relative to a long-term baseline (McKee et al., 1993). We adopted the UK gridded SPI data set of (Tanguy et al., 2017), which used 1961–2010 as its reference period. The SPI can be calculated for different accumulation periods: for example, SPI1 reflects precipitation anomalies over a 1 month window, while SPI12 reflects anomalies accumulated over 12 months. For each catchment, we defined winter SPI1 as the mean of the monthly SPI1 values for December–February, and annual SPI12 as the September SPI12 value for each water year. Winter SPI1 was used because winter precipitation is the main driver of hydrological connectivity in UK catchments, when soils approach saturation and both surface and subsurface pathways become activated (Muchan et al., 2015). In contrast, summer flows are controlled largely by evapotranspiration and accumulated soil-moisture deficits (Kilsby et al., 2019), making SPI12 a more appropriate indicator of antecedent conditions relevant to low-flow behaviour. Population density was derived from the UK gridded population 2011 based on Census 2011 published by Environmental Information Data Centre (Reis et al., 2017). The density of Wastewater Treatment Plant (WWTPs) was calculated by dividing the number of WWTPs in a catchment by the catchment area Department for Environment, Food and Rural Affairs (2020).

Daily concentrations were reconstructed using Weighted Regression on Time, Discharge and Season (WRTDS), which was implemented in the R package EGRET (version 3.0.9) (Hirsch et al., 2023, 2010). This approach estimates daily concentrations from irregularly sampled data using a locally weighted regression: ln⁡(Ci)=β0,i+β1,iti+β2,iln⁡(Qi)+β3,isin⁡(2πti)+β4,icos⁡(2πti)+εi where ti is time in decimal years, Ci is the concentration on day i, Qi is the daily discharge, β0 is the intercept, β1 captures long-term concentration trends, β2 represents the local C–Q slope, and β3 and β4 account for seasonal cycles, and εi is the residual error. These coefficients are fitted through regression at each time point, weighting observations by similarity in time, discharge, and season. The WRTDS method estimates a smooth concentration surface in time–discharge–season space by performing a locally weighted regression for every modelled day, producing thousands of smoothly varying coefficient vectors for each catchment rather than a single global fit. The local weighting is defined by three half-window widths: 7 years in time, 0.5 years in season, and 2 natural-log units in discharge. This dynamic approach allows the concentration–discharge relationship to vary smoothly over time, reduces bias from irregular sampling (especially under-representation of high flows), and handles censored values effectively (Hirsch and De Cicco, 2015). We further used the locally estimated β2 coefficients to classify each catchment's export regime as dilution (β2<-0.1), chemostasis (-0.1≤β2≤0.1), and mobilisation (β2>0.1) following established thresholds (Zhang, 2018; Herndon et al., 2015). The R scripts published by (Zhang et al., 2016) were used to estimate and extract the β2 coefficients. To support seasonal synchrony analysis, reconstructed daily concentrations were aggregated to monthly mean values. This produced a more complete and temporally consistent representation of nitrate dynamics than the observed data alone and enabled robust comparisons across catchments and years. For the synchrony analysis, both reconstructed concentrations and observed discharge were subsequently aggregated to monthly mean values, and all synchrony classifications were determined at the monthly scale based on the timing of monthly maxima and minima. To assess the adequacy of WRTDS-estimated concentrations for identifying monthly peaks, we computed the temporal Spearman correlation between observed and modelled monthly concentrations to assess agreement in temporal patterns relevant to the synchrony analysis. Across sites, spearman correlations were calculated showing a median ρ of 0.72 (IQR=0.134). No sites were excluded based on model performance.

We confirmed that both discharge and nitrate concentration exhibit a clear seasonal peak–trough structure at all sites, based on their long-term monthly means. We then identified, for each site and each year, the months of maximum and minimum discharge, as well as the month of the maximum nitrate concentration. Based on the consistency between these seasonal timings, we first defined the annual synchrony status for each year at each site as: (1) QMax-Synced when the month of maximum concentration coincided with or fell within one month (±1 month) of the maximum discharge month, (2) QMin-Synced using the same criteria as for QMax-Synced, except applied to the minimum discharge month, or (3) Asynced for years/catchments that met the criteria for neither category. This window was chosen to capture seasonal synchrony while avoiding over-interpretation of exact monthly peak timing from irregular observations and WRTDS-smoothed daily estimates. A broader window would blur seasonal contrasts. We then used these annual synchrony categories in two ways throughout the analysis. First, as a catchment level classification, where each catchment was assigned a dominant type (QMax-Synced, QMin-Synced) if that category occurred in more than 50 % of years, and all remaining catchments were classified as Asynced. Second, we assessed the consistency of synchrony within each catchment across years, using the proportion of years in which it exhibited QMax- or QMin-synchrony as a measure of interannual synchrony variability. This dual approach allowed us to characterise both the dominant synchrony pattern at each site and its temporal stability (or lack of) over two decades of record. One catchment (NW-88003442) exhibited a clear structural change in the early 2000s. As a robustness check, we recalculated its synchrony using only post-2004 data; the dominant synchrony category remained unchanged, so the full record was retained for consistency across sites.

To analyse the spatial variability of nitrate-discharge synchrony and its controls, the mean discharge and nitrate concentration for each catchment were calculated. The ratio of the coefficients of variation of discharge (CV_q) and concentrations (CV_c) was also calculated to quantify the relative hydrological modulation of nitrate variability.

To understand the catchment controls on the two synchrony patterns, a series of catchment descriptors (Table 1) were selected, and a Random Forest (RF) model was applied to relate dominant synchrony type (QMax-Synced vs QMin-Synced) to catchment characteristics. RF analysis was chosen for its robustness and ability to handle complex interactions within the data (Breiman, 2001). Asynced catchments were excluded from this analysis because they do not exhibit a dominant synchrony pattern, as such including them would reduce interpretability of contrasts between the two synchronous modes. A Random Forest classifier was implemented using the ranger (Version 0.16.0) engine via mlr3 (Version 0.22.1) in R (Lang et al., 2019; Wright and Ziegler, 2017). No hyperparameter tuning was performed; ranger default settings were used except that the number of trees was set to 500. A three-repeated 10-fold cross-validation procedure was used to evaluate model performance and its generalizability. Following cross-validation, a final RF model was trained on the full dataset, and permutation-based variable importance values were extracted from the underlying ranger model, following Altmann et al. (2010). Empirical permutation p-values (<0.05) were obtained from 1000 random permutations of the response variable to identify descriptors whose importance exceeded the null distribution. Variables identified as important by the RF model were tested for significant differences between synchrony types using the non-parametric Wilcoxon rank-sum test (R Core Team, 2024). Given the modest sample size and uneven class distribution, our aim was to use the RF primarily for variable importance ranking rather than formal prediction. To test the possible internal heterogeneity of QMin-Synced catchments, we also calculated Spearman correlations among land-use variables and two nitrate metrics (median concentrations and CVc/CVq) to assess whether different nitrate metrics respond differently to land use within QMin-Synced catchments.

In addition, we assessed whether climatic variability and catchment properties influenced the frequency with which a catchment exhibits QMax, QMin, or Asynced behaviour from year to year. Initially we tested whether climate anomalies, represented by winter SPI1 and annual SPI12, differed between Synchronous (QMax- or QMin-Synced) and Asynced years within each synchrony pattern using paired Wilcoxon signed-rank tests. Second, we assessed the climate–discharge linkage by correlating SPI metrics with the annual maximum and minimum of the monthly mean discharge (hereafter MaxQ and MinQ), to evaluate how climatic wetness anomalies were expressed in discharge extremes. Then, to test whether discharge extremes were directly associated with synchrony outcomes, we converted annual MaxQ and MinQ to percentile ranks relative to each site's full distribution of monthly discharges. We then compared MaxQ percentiles between QMax- and non-QMax years in catchments where QMax-synchrony was more frequent and compared MinQ percentiles between QMin- and non-QMin years in catchments where QMin-synchrony was more frequent. Finally, we assessed relationships between spatial synchrony variability, defined as differences among catchments in their long-term tendency toward QMax-, QMin-, or Asynced behaviour, and potential controlling factors using Spearman rank correlations and Wilcoxon rank-sum tests. Analyses were conducted in Python using the numpy, pandas, seaborn, matplotlib and scipy packages (Virtanen et al., 2020; Reback et al., 2020; Harris et al., 2020; Waskom, 2021; Hunter, 2007).

Catchment synchrony classifications and their temporal stability differed across the 66 catchments (Figs. 1 and 2). QMax-Synced catchments (28.8 %) were primarily located in southern and southwestern England, in these catchments nitrate peaks typically aligned with winter high flows. QMin-Synced catchments (25.8 %) were concentrated in the northwest, showing peak nitrate concentrations during or near summer low flows. The Asynced catchments (45.5 %) were more broadly distributed across the country and did not exhibit a consistent seasonal alignment between nitrate and discharge.

QMax-Synced sites exhibited clear winter nitrate peaks (Fig. 3a), with highest median concentrations in February (6.45 mgL-1, IQR: 4.19–6.95 mgL-1) and lowest in September (4.39 mgL-1, IQR: 2.80–5.70 mgL-1). In contrast, QMin-Synced catchments typically showed a reversed seasonal pattern, with lowest concentrations during winter high flow (January: 4.40 mgL-1, IQR: 3.12–4.95 mgL-1) and peak concentrations during summer low flow (July: 6.15 mgL-1, IQR: 4.15–8.02 mgL-1; Fig. 3b). Asynced sites displayed flatter, more spatially variable nitrate regimes, without a dominant seasonal signal. While individual sites may have distinct seasonal patterns, the lack of alignment in their timing and magnitude resulted in the group-average curve appearing flat. Despite these marked seasonal differences, median nitrate concentrations did not differ significantly among the three synchrony types (Fig. S2). Overall, interannual variability in nitrate concentrations was substantially lower than variability in discharge (median CVc/CVq=0.19, IQR: 0.14–0.32). However, QMin-Synced catchments showed significantly higher CVc/CVq ratios compared to the other two synchrony types (Fig. S3), indicating stronger hydrological modulation of nitrate variability in these catchments. Descriptive statistics for discharge, nitrate concentration, and other related variables across all catchments are summarised in Table S1.

Although our classification groups catchments based on their dominant synchrony type, catchments also exhibited varying degrees of interannual variation in their synchrony status (Fig. 2). Among QMax-Synced sites, the proportion of QMax-aligned years ranged from 58 %–90 %; similar ranges were found for QMin-Synced sites. Asynced catchments showed greater fluctuation between synchrony types. Consistent with this interannual variability, synchronous or asynced years are sparsely distributed through time (Fig. S4), with no indication of a long-term trend.

Catchments within each synchrony classification exhibited contrasting nitrate export behaviours, as reflected in the relationship between discharge and annual peak nitrate concentrations. Here, the interannual slope refers to the regression slope between annual peak nitrate concentrations and the discharge during the month of peak concentration (in log units) for each catchment (Fig. 4a–c). QMax-Synced catchments showed shallower and more variable interannual slopes that were typically weakly positive or near-zero. Within this group, 42.1 % of catchments exhibited positive slopes, and 10.5 % showed negative slopes. β2 values (i.e. C–Q slopes estimated from the WRTDS model) for QMax-Synced catchments fell mostly in the chemostasis to mobilisation range (Mean β2: 0.07±0.10).

In contrast, QMin-Synced catchments showed the strongest inverse coupling to discharge with consistently steep negative slopes between discharge and peak nitrate, and correspondingly negative β2 values (Mean β2: -0.42±0.19). Within this group, almost 58.8 % of catchments had a negative slope between peak nitrate concentrations and corresponding log(discharge). Asynced catchments also exhibited predominantly negative Q–C slopes (Mean: -0.12±0.12), but with greater scatter and a broader mix of β2 values spanning dilution and chemostasis regimes. Nevertheless, many catchments exhibited negative β2 values, suggesting that their nitrate peaks are often shaped by dilution-like behaviour despite lacking a consistent seasonal alignment with flow. Despite these differences in export dynamics, peak nitrate concentrations did not differ significantly among synchrony types (Kruskal–Wallis's test: H=3.62, df=2, p=0.16).

Secondly, we examined the interannual stability of the timing of nitrate and discharge peaks from year to year. For each synchrony type, we analysed the consecutive-year change in the month of peak nitrate concentration and compared it with the corresponding change in the month of maximum or minimum discharge (Fig. 4d and e). In QMax-Synced catchments (Fig. 4d), both nitrate and discharge peak timing intervals were mostly stable around the 1 : 1 line, with most intervals confined within ±1 month. This temporal coherence is consistent with relatively stable winter-dominated nitrate delivery. In QMin-Synced catchments (Fig. 4e), the interval between minimum-flow months varied more widely across years and seasons, yet the corresponding intervals between nitrate peaks tracked these changes closely.

To understand which catchment attributes best explain differences in the dominant synchrony behaviour identified in Sect. 3.2, we used a Random Forest classification model to relate land use, hydrology, and landscape properties to synchrony type (QMax-Synced, QMin-Synced). The random forest classification model achieved a classification error of 12.2 %, corresponding to an overall accuracy of 87.8 %, indicating strong separability between the two synchrony types given the selected descriptors. This indicates that the model was able to distinguish between the synchronous patterns based on the selected predictors. After removing descriptors whose importance was not statistically significant, the classification error increased only marginally to 12.5 % (+0.3%).

The most influential catchment descriptors identified by the RF model are consistent with known hydrological and land use controls on nitrate dynamics (Fig. 5). Urban Land cover ranked as the most influential descriptor, followed by the Standard Percentage Runoff (SPR), Arable Land, Baseflow Index (BFI) and the percentage of High Permeable Surface. Random Forest variable importance scores are shown in Fig. S5. QMax-Synced catchments had a significantly smaller percentage of urban areas while QMin-Synced catchments had the highest (p<0.01). In contrast, the percentage of arable land was significantly higher in QMax-Synced catchments compared to QMin-Synced ones. Regarding the hydrological characteristics, linear regression analysis indicated a strong inverse relationship between SPR and BFI, with an R2 value of 0.71, indicating, not surprisingly, that these variables capture overlapping aspects of catchment hydrological behaviour. QMax-Synced catchments exhibited lower SPR values and higher BFI. The proportion of highly permeable surface deposits, reflecting infiltration capacity and groundwater recharge potential, was lowest in QMax-Synced catchments. Although catchment area did not reach statistical significance (p=0.07), QMax-Synced catchments tended to have smaller catchment areas than QMin-Synced catchments. Within QMin-Synced catchments, median nitrate concentrations showed positive (though non-significant) associations with arable and urban land cover, and a significant negative association with grassland cover (ρ=-0.50, p=0.04). In contrast, CVc/CVq was most strongly correlated with urban land (ρ=0.55, p<0.05), population density (ρ=0.52, p<0.05), arable land (ρ=-0.59, p<0.05) and woodland cover (ρ=-0.61, p<0.01). All the other correlation results are shown in Fig. S6.

While Sect. 3.3 identified catchment characteristics that explain dominant synchrony types across space, here we examine (i) what controls interannual variability in synchrony within catchments, and (ii) how long-term variability in synchrony composition relates to catchment attributes. For QMax-Synced catchments, winters were significantly wetter during QMax-Synced years than Asynced years (mean SPI1 difference, Δ=0.27±0.32, p<0.01, Table 2). No significant differences in SPI were observed between QMin-Synced and Asynced years in QMin-Synced catchments (p>0.05). In catchments classified as Asynced overall, the years identified as QMin-Synced were significantly wetter, both in terms of winter SPI1 and annual SPI12, than years identified as either QMax-Synced or Asynced (p<0.01 for both comparisons). Across the 56 sites, both winter SPI1 and SPI12 showed strong positive correlations with the annual maximum of monthly mean discharge (MaxQ), with median ρ values of 0.78 and 0.78, respectively, and 49 and 56 sites showing statistically significant relationships. In addition, SPI12 was also positively correlated with the annual minimum of monthly mean discharge (MinQ) (median ρ=0.51; 35 sites significant). In catchments where QMax-synchrony occurred more frequently (36 sites), QMax years did not show significantly higher percentiles of maximum monthly discharge than non-QMax years (median 0.897 vs. 0.910, p=0.30). In contrast, in catchments where QMin-synchrony was more frequent (30 sites), non-QMin years showed slightly higher percentiles of the minimum-flow month than QMin years (median 0.086 vs. 0.068). Although the difference is modest, it is statistically distinguishable (p=0.007).

In addition to the interannual variability, we also considered spatial variability in synchrony composition captured as long-term differences in the relative occurrence of QMax-, QMin- and Asynced years across catchments. The ternary plots (Fig. 6) summarise the long-term synchrony composition at each site by showing the relative proportion of QMax-, QMin- and Asynced years. In these diagrams, each vertex represents 100 % dominance of one synchrony type, with positions along the edges indicating a mixture of two types, and points near the centre represent a balanced mixture of all three. The colour scale applied to each point represents values for a selected catchment descriptor (e.g., Arable land percentage, Drainage Path Slope), and visualises how synchrony composition varies along environmental gradients. These colour gradients are consistent with the spatial correlations shown in Fig. S7. No catchments were located at a single vertex, instead, most exhibited a mix of synchrony types, with Asynced catchments spanning a particularly wide range. No sites are located near both the QMax and QMin vertices simultaneously, indicating that switching occurs mainly between a dominant mode and Asynced behaviour rather than directly between the two synchronous modes. These catchment-level synchrony metrics were then related to catchment attributes to identify the key controls on the dominant synchrony state (i.e. providing insight into why certain catchments display QMax-, QMin- or mixed synchrony). Within QMax-Synced catchments, a higher share of QMax years was associated with greater arable land cover (ρ=0.58, p<0.05), lower Drainage Path Slope (DPS; ρ=-0.50, p<0.05) and low standard percentage runoff (SPR, ρ=-0.53, p<0.05). In contrast, in QMin-Synced catchments, urban land cover (ρ=0.53, p<0.05), population density (ρ=0.58, p<0.05), as well as higher CVc/CVq and the Proportion of Time Soils Are Wet (PROPWET) (both ρ=0.58, p<0.05), were all positively associated with the proportion of QMin-Synced years. Asynced catchments behaved as transitional systems with a higher density of WWTPs shifting the year mix towards QMin-Synced years (ρ=0.36, p<0.05) and away from QMax-Synced years (ρ=-0.50, p<0.05). A higher baseflow index (BFI) tended to support QMax-Synced years (ρ=0.38, p<0.05), while larger catchment area (ρ=0.39, p<0.05), and lower SPR (ρ=-0.38, p<0.05) were linked to a greater prevalence of QMin-Synced years. A full summary of additional correlations is provided in Fig. S7.

The emergence of three synchrony regimes demonstrates that space and time imprint differently on peak nitrate export. We therefore interpret synchrony primarily through catchment-scale source–pathway and dilution–mobilisation controls, while recognising that unmeasured in-channel or impoundment processes may contribute to residual variability in nitrate peak timing. Each regime reflects a distinct organisation of sources, pathways, and hydrological connectivity that together determine when nitrate and flow peaks coincide or diverge. Spatially fixed contrasts, driven by land use, geology, and drainage infrastructure, provide a first order control to the relatively stable QMax- and QMin-synchrony types. Asynced catchments, in contrast, lack a single dominant source–pathway configuration, resulting in a mixed synchrony signature. In this sense, synchrony is not a static property but an emergent expression of how catchment structure interacts with climatic forcing. The following sections examine each synchrony type in turn, tracing how characteristic source–pathway–connectivity configurations lead to QMax, QMin, or Asynced behaviour.

Our peak-based analysis showed that, although both nitrate concentration and discharge follow a consistent seasonal cycle on average, the timing of their annual peaks varies substantially among years and among catchments. Earlier studies have shown that riverine nitrate concentrations generally track discharge seasonality (Ebeling et al., 2021; Van Meter et al., 2019). Our results extend this understanding by showing that synchrony itself fluctuates across time and space. The strength and timing of nitrate and flow coupling vary with hydro-climatic conditions and catchment characteristics. This variability reveals how climate sets the potential for synchrony, while local land use and hydrological structure determine whether that potential is realised.

Asynced behaviour was characterised by a lack of consistent alignment between peak nitrate concentrations and discharge. These asynced catchments are common (45.5 % of study basins), broadly distributed across England and display weak seasonal patterns in monthly nitrate concentrations. They occupy a continuum shaped by the interplay of multiple nitrate sources and flow pathways. Their intermediate export behaviour reflects neither persistent mobilisation nor strong dilution dominance, indicating mixed source contributions rather than a single controlling process. Similar Asynced N patterns in western France have been attributed to legacy nitrogen stores in groundwater, bottom-loaded nitrate in soil profiles, and spatially variable hydrological connectivity together maintained persistently high but weakly seasonal concentrations (Guillemot et al., 2021; Abbott et al., 2018). Moreover, anthropogenic activities (e.g., tile drainage, groundwater extraction, and land use changes) have also been found to amplify chemical contrasts between shallow and deep flow paths, further decoupling nitrate responses from surface hydrology and promoting asynchrony (Zhi and Li, 2020).

In most Asynced catchments, all three synchrony states were observed across years, highlighting their mixed and shifting controls. This instability likely reflects the absence of persistent hydrological forcing. Whereas QMax-synchrony requires enhanced winter connectivity and QMin-synchrony requires sufficiently severe low-flow conditions, Asynced catchments often experience neither consistently. Asynced catchments, likely lacking either consistent high-flow connectivity or prolonged low-flow extremes, are therefore highly prone to switching between synchrony states. Unlike consistently QMin-Synced catchments, where continuous anthropogenic loading and extreme low flows dominated and where SPI showed little influence, Asynced catchments were more sensitive to climatic anomalies. Wetter winters (higher SPI1) and wetter years (high SPI12) occasionally promoted QMin-like behaviour, though not through urban effluents but likely because increased recharge temporarily reconnected shallow groundwater or legacy nitrate stores to the stream network. These transient connections can mimic the timing of QMin behaviour, producing nitrate peaks during low-flow periods. However, they likely arise from hydroclimatic modulation of connectivity rather than from persistent point-source dominance, and represent the primary context in which legacy nitrate appears to contribute to synchrony.

Beyond hydroclimatic influences, nitrate export regimes in Asynced catchments were further shaped by catchment size and anthropogenic pressures. Larger catchments tended to dilute diffuse inputs, favouring QMin-like behaviour, while denser wastewater infrastructure reinforced nitrate peaks during low flow periods. Year-to-year differences were also likely influenced by broader environmental variability, such as air temperature, extreme rainfall, or antecedent soil moisture, that can alter uptake and denitrification efficiency and thus accentuate asynchrony (Van Meter et al., 2019). Such effects may be most plausibly expressed in Asynced catchments, where no single source–pathway configuration consistently dominates and where longer travel times, impoundments, or variable low-flow residence conditions could allow in-channel processing to modify peak timing. Importantly, because none of these controls acts consistently across years, correlations with individual drivers remain modest. These results indicate that Asynced catchments are not characterised by greater climatic or hydrological variability per se, but by greater sensitivity to it. With weaker structural constraints than the spatially organised QMax and QMin regimes, their synchrony state shifts readily in response to interannual changes in wetness, storage, or loading. This heightened sensitivity may arise because Asynced catchments lack a single dominant source–pathway configuration, as diffuse, groundwater, and urban inputs likely all contribute, but their relative influence depends on hydrological thresholds that vary from year to year. Testing these threshold mechanisms would require time-varying data not consistently available across the national dataset, including event-scale rainfall and discharge, soil-moisture and groundwater dynamics, drainage infrastructure, WWTP discharge chemistry, and in-stream indicators such as residence time or denitrification. When those thresholds are reached, even modest climatic anomalies can switch the dominant transport pathway, altering whether export resembles QMax- or QMin-like behaviour. The fact that Asynced catchments constitute the most common catchment type in our dataset underscores the importance of this temporally responsive regime, one that is easily overlooked in analyses focused solely on spatial contrasts in land use or source dominance. Together, these results show that the synchrony framework adds a critical temporal dimension to understanding nitrate–flow coupling, revealing how small climatic or infrastructural perturbations can reorganise export dynamics across seasons and years.