Decoupling factors influencing spatial and temporal trends of total nitrogen and ammonia nitrogen levels in lakes across China

Shi, Chenyi; Wang, Xihua; Xu, Y. Jun; Dai, Chaomeng; Zhou, Nianqing; Fu, Rongbing; Luo, Chengming; Mao, Boyang; Jia, Shunqing; Lv, Qinya; Liu, Zejun; Ji, Xuming; Dai, Yan; Rong, Yanxin

doi:10.5194/egusphere-2026-3574

Preprints

https://doi.org/10.5194/egusphere-2026-3574

Preprints

06 Jul 2026

| 06 Jul 2026

Status: this preprint is open for discussion and under review for Hydrology and Earth System Sciences (HESS).

Decoupling factors influencing spatial and temporal trends of total nitrogen and ammonia nitrogen levels in lakes across China

Chenyi Shi, Xihua Wang, Y. Jun Xu, Chaomeng Dai, Nianqing Zhou, Rongbing Fu, Chengming Luo, Boyang Mao, Shunqing Jia, Qinya Lv, Zejun Liu, Xuming Ji, Yan Dai, and Yanxin Rong

Abstract. The heterogeneity of lake nitrogen pollution in its spatial distribution and long-term evolution influences ecosystem functioning and the effectiveness of water environment management. However, whether the drivers shaping the spatial patterns of different nitrogen forms differ from those governing their temporal dynamics remains poorly understood at large spatial scales. Here, we constructed monthly time series (2010–2024) of total nitrogen (TN) and ammonia nitrogen (NH₃-N) concentrations for 3,020 lakes across China’s five limnological regions using an ensemble machine learning framework and to reveal the difference in factors that influence spatial patterns and long-term changes of the two nitrogen forms at the large scale. The results showed that the factors determining where lakes are nitrogen-enriched are not necessarily those controlling whether nitrogen conditions improve or deteriorate over time and that different nitrogen forms respond to management and environmental change through distinct pathways. For TN, the spatial patterns are jointly shaped by external nitrogen loading, land cover, and hydro-climatic conditions, highlighting strong landscape-scale controls. However, its long-term evolution is governed more strongly by temporal variations in external nitrogen inputs than by static spatial characteristics. In contrast, NH₃-N exhibits a more direct and rapid response to external emission reduction measures, with both its spatial and temporal dynamics demonstrating high sensitivity to changes in anthropogenic nitrogen inputs. By highlighting the decoupled controls on spatial patterns and temporal trends, this study underscores the necessity of shifting lake management from static, location-based regulation toward an integrated 'state-rate' adaptive strategy.

Received: 18 Jun 2026 – Discussion started: 06 Jul 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 3619 KB)

Supplement (141 KB)

Download & links

Chenyi Shi, Xihua Wang, Y. Jun Xu, Chaomeng Dai, Nianqing Zhou, Rongbing Fu, Chengming Luo, Boyang Mao, Shunqing Jia, Qinya Lv, Zejun Liu, Xuming Ji, Yan Dai, and Yanxin Rong

Status: open (until 07 Oct 2026)

Post a comment Subscribe to comment alert

CC1:
'Comment on egusphere-2026-3574', Michael McClain, 31 Aug 2026 reply
This manuscript presents a monthly analysis of TN and NH₃-N concentrations in 3020 lakes across China and their spatial and temporal relationship to select influential factors over a 15-year period (2010-2024). By applying an ensemble machine learning framework, the authors assessed the relative strength of different factors in predicting the spatial and temporal patterns in TN and NH₃-N concentrations. The main findings are that, depending on the limnological region of the country, external nitrogen loading, hydro-climatic dynamics, soil N & land cover, and topography structure each appeared as dominant factors in predicting the spatial variability of TN, while only external nitrogen loading and soil N & land cover appeared as dominant factors in predicting the spatial variability of NH₃-N. A similar distribution of dominant factors was found with respect to trends in TN over the 15-year study period, while NH₃-N trends were related predominantly to external N loading. Differences between the relationships with factors were attributed to the influence of legacy nitrogen stored in the system on TN concentrations and the higher reactivity of NH₃-N concentrations in response to continued external N loading. The findings led the authors to recommend nitrogen-form-specific management strategies to account for the differing influences.
In reviewing this manuscript, I was not able to assess the ensemble machine learning framework applied, as it is outside my expertise. My review therefore focuses primarily on data used, the interpretation of the results and the recommendations.
Overall, I found the manuscript to present a thorough and comprehensive analysis of a remarkable dataset. The results are significant and of value, at least with respect to Total N, in addressing problems of N contamination across China and more broadly. It is also well written.
My primary concern centres around the use of NH₃-N as a response variable for such a national and catchment scale analysis and the influence this has on the final conclusions. The analysis infers cause and effect relationships between the predictor variables (e.g. external N loading) and the response variables (TN and NH₃-N concentrations). This leads to the following statement in the Conclusions section “Based on these findings, we postulate that both the spatial distribution and long-term trend of lake TN across a large geographical region are reflected by watershed-scale cumulation of nitrogen, and that the long-term trend of lake NH₃-N is primarily affected by external nitrogen loading”. From this the authors go on to state “The different responses of lake nitrogen forms in spatial and temporal characteristic across China highlight the need for nitrogen-form-specific management strategies.”
While I judge the analysis and eventual conclusions related to TN concentrations to be reasonable, I do not feel the same about those for NH₃-N. Ammonia (NH₃) is a highly reactive compound of nitrogen in natural waters generated during the ammonification of organic nitrogen and generally quickly converted to ammonium (NH4⁺), oxidised to nitrate (NO₃^-), or taken up and assimilated by plants. The predictor variables considered in this study are quite distal from the conditions in the lakes themselves. NH₃-N concentrations could be much better predicted using TN concentrations in the lake, pH and dissolved O₂ concentrations. Nitrate would have been the better dissolved inorganic form of N to use for the analysis and may have even resulted in similar conclusions, but I understand that NH₃-N was the only option available in the CNEMC dataset. CNEMC monitoring of NH₃-N is understandable because it is a priority pollutant and serves as a rapid warning indicator of failures in water treatment systems or flushing events from nearby contamination sources (like feedlots). It is valuable for detecting local effects, but it is not a good indicator of catchment scale characteristics.
The poor utility of NH₃-N concentrations as an indicator of external N loading is evident in the concentrations reported in the manuscript. The observed mean concentrations reported in lines 336-339 (maximum 0.159 mg/L) are low and acceptable for water quality Classes I and II in China. By comparison the national multi-year mean TN concentration of 2.07 mg/L in the lakes is slightly above what is acceptable even for water quality Class V. So, comparatively, TN concentrations reflect highly enriched conditions while NH₃-N concentrations do not. In short, mean regional TN concentrations in the lakes are a water quality problem, while mean NH₃-N concentrations are not.
TN concentrations may very well be influenced by external N loading in the catchment, reflecting as the authors state “the widespread inertia inherent in watershed- and lake-scale nitrogen pollution and recovery processes at the global scale”. But NH₃-N concentrations are likely only influenced by the TN concentrations in the lake, which when combined with pH and dissolved oxygen conditions in the lake are not so high as to result in widespread NH₃-N concentrations exceeding national water quality standards.
I believe these considerations may be so significant as to remove NH₃-N from the analysis. If the authors choose to keep NH₃-N in the analysis, I recommend that the considerations raised above be reflected in the revision of the manuscript. TN and NH₃-N should not be assumed to respond in equal strength to the variations in the predictor variables distal to the river. The limitations of using NH₃-N for such a study should be acknowledged. Results and conclusions related to NH₃-N should also therefore be qualified.
Other points for consideration
Throughout the manuscript the authors describe TN of NH₃-N spatial patterns or trends to be “attributed to”, “governed by”, “affected by”, “influenced by”, etc., the variables representing drivers. For example, lines 461-462 say “the long-term changes in NH₃-N concentrations were primarily governed by external nitrogen loading”. This leads to the clear inference of the direct cause-and-effect relationships I referred to above. But the associations between N concentrations and variables used in the analysis are not causal relationships… they are predictive statistical relationships. This language should be revised throughout the manuscript, replacing words that imply causality with correct representations such as “associated with”, “statistically related to”, “predictive of”, etc. This change in language will partially, but not fully, address the concerns raised above about the use of NH₃-N.

Lines 66-67: The text states “Previous studies have documented the differential responses of nitrogen forms to climatic, hydrological, and land-use changes at lake or watershed scales”. Careful in claiming that studies have already "documented" the responses of different N forms to climate change. Your citations here do not support that. Du et al. (2023) investigated organic matter with no analysis of N forms, and Jiang et al. (2026) report modeling results of future conditions. If you want to keep the statement, your references should document studies of N responses to climate change that have already occurred.

Lines 107-110: The sentence “This highly compressed spatiotemporal heterogeneity makes it difficult for analyses based on single catchments to systematically reveal the divergent drivers of different nitrogen forms across spatial states and long-term evolutionary processes” can be deleted.

Section 2.1 Study Area: The first sentence says the study analysed 3020 lakes, but then in line 131 the distribution of study lakes across the five limnological regions is presented as “483, 690, 153, 1, 206, and 488”. These numbers sum to only 2021 lakes, with only 1 indicated in the EPL. Why the difference?

Lines 428-429: Large seasonal variability is indicated in external nitrogen loading, but nitrogen loading data in Table 1 are annual. I was not able to find the description of how the annual data were distributed seasonally? Is this described somewhere? If not it needs to be added.

Minor points
Line 133: Fig 1b is cited, but I do not see a 1b in the figure.
Line 166: “(ERA5)” should be (ERA5-Land).
Lines 515-519: Citation needs to be added for decrease in N loads from 6.8 to 3.5 million tons. What is time unit… per yr?

Reply
Citation: https://doi.org/10.5194/egusphere-2026-3574-CC1
- AC1: 'Reply on CC1', Xihua Wang, 07 Sep 2026 reply
  
  We would like to sincerely thank the reviewer for his/her supporting and for taking the time to review our manuscript. Your good suggestions have increased our papers quality. thank you very much! In this reply, we have copied the comments in black. Our responses are entered in blue. You can find the responses in the attached file.
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-3574-AC1

Chenyi Shi, Xihua Wang, Y. Jun Xu, Chaomeng Dai, Nianqing Zhou, Rongbing Fu, Chengming Luo, Boyang Mao, Shunqing Jia, Qinya Lv, Zejun Liu, Xuming Ji, Yan Dai, and Yanxin Rong

Supplement

https://doi.org/10.5194/egusphere-2026-3574-supplement

Chenyi Shi, Xihua Wang, Y. Jun Xu, Chaomeng Dai, Nianqing Zhou, Rongbing Fu, Chengming Luo, Boyang Mao, Shunqing Jia, Qinya Lv, Zejun Liu, Xuming Ji, Yan Dai, and Yanxin Rong

Viewed

Total article views: 250 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
191	44	15	250	18	12	12

HTML: 191
PDF: 44
XML: 15
Total: 250
Supplement: 18
BibTeX: 12
EndNote: 12

Views and downloads (calculated since 06 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	128	19	9	156
Aug 2026	38	18	3	59
Sep 2026	25	7	3	35

Cumulative views and downloads (calculated since 06 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	128	19	9	156
Aug 2026	38	18	3	59
Sep 2026	25	7	3	35

Viewed (geographical distribution)

Total article views: 197 (including HTML, PDF, and XML) Thereof 197 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 07 Sep 2026

Short summary

We studied how nitrogen pollution changed in 3,020 lakes across China from 2010 to 2024 using monitoring records, environmental data, and machine learning. We found that lakes with high nitrogen today are not always the ones getting worse over time. Total nitrogen changed slowly and was shaped by both climate and long-term pollution buildup, while ammonia nitrogen responded rapidly to pollution control. These results show lake management should target both current pollution and future change.


Total:	0
HTML:	0
PDF:	0
XML:	0