the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Feb 2026
Riverine phosphorus gain and loss across the conterminous United States
Abstract. Excess riverine phosphorus represents a preeminent catalyst for water quality degradation. Spatial mapping and characterization of the net gain and loss of riverine phosphorus help discern the critical source areas. Here, we developed a dataset encompassing phosphate (PO43-) and total phosphorus (TP) gain and loss across catchments in the conterminous United States (CONUS). We compiled 51,394 PO43- and 285,675 TP concentration data points and estimated PO43- and TP loads at 963 and 2,317 stations, respectively. Next, we leveraged the upstream-downstream topology information from the National Hydrography Dataset Plus (NHDPlus) catchment map at the Hydrologic Unit Catalogue-12 (HUC12) level to derive the net gain and loss of riverine phosphorus across catchments in the CONUS. Such maps can be used to estimate potential contributions of point and non-point sources to riverine phosphorus pollution at refined spatial scales, identify different major factors controlling local riverine P gain and loss compared to P loads, and evaluate watershed model’s fidelity for representing riverine P cycling. The resultant dataset is provided in Excel (.xlsx) format, accessible at Figshare (https://doi.org/10.6084/m9.figshare.28509317, Wang et al., 2025). Leveraging the HUC12 information for spatialization, the new datasets aim to address the existing gap in regional characterization of riverine phosphorus and support effective management practices across the CONUS.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Earth System Science Data.
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.- Preprint
(2135 KB) - Metadata XML
-
Supplement
(1330 KB) - BibTeX
- EndNote
Status: open (until 20 Apr 2026)
- RC1: 'Comment on essd-2025-743', Anonymous Referee #1, 02 Apr 2026 reply
-
RC2: 'Comment on essd-2025-743', Anonymous Referee #2, 13 Apr 2026
reply
Review of ESSD-2025-743 “Riverine phosphorus gain and loss across the conterminous United States” by Wang et al.
Phosphate and TP loads were estimated at 963 and 2,317 stations across the CONUS. Loads were linked to NHD to apply topology routing and estimate riverine net gains and losses. This reviewer finds immediate value in the dataset and initial gain and loss estimates. Comparison of Loadest to WRTDS is value added. This dataset should serve as a foundation for other modeling and analyses. Well done! This reviewer has only minor comments for the authors to consider to perhaps further clarify the presentation.
General comments and questions:
What is your time period? Was it recent 2015-2020? That info is super important towards interpretation. It should be stated in the figshare dataset description too. And perhaps even included in the title.
The Skinner and Maupin, 2019 is your dataset for point sources? This is a good comprehensive dataset. So that means point source influence was represented at the annual timescale? A reader may benefit from that knowledge.
The discussion of general factors is helpful. Is there a focus to determine or assess change or trends?
Results displayed by region is effective. Summarizing by HUC12 will also be immediately useful to others.
Model diagnostics seem reasonable. A statement regarding why there are possible differences between Loadest and WRTDS may benefit the reader.
How was “riverine removal” estimated? That is not clear to this reviewer. Is it when upstream to downstream is negative? Or did you just assume 12% loss? If yes, that would assume that reservoirs/lakes are mostly responsible for removal. And were reservoirs/lakes considered in this loss calculation and interpretation?
Figure 6: There is useful information here, but it seems worth suggesting something like a dynamic watershed model (e.g., SPARROW) as a possible next step for another approach to gains and losses. This reviewer agrees that this dataset is useful in supporting evaluation and diagnosis of watershed models, and because you are offering an initial cut, it may be worth making a more explicit recommendation.
Did streamflow only come from NWIS? Did you use only USGS gages? You used more than the GagesII reference gages, correct? If only reference gages, linking to agriculture is limited. And how were those paired with your discrete P data? Did you only consider stations with P and streamflow data, or did you collect P data still close to a streamflow gage? A concise statement is needed for the reader.
Is the WQP pull comprehensive, or are there still possible holes or gaps in that data pull?
Specific comments:
27: Eutrophication of what? A Wurtsbaugh citation implies lakes. Perhaps add “inland waters and estuaries” to benefit the reader. This reviewer agrees there should be more focus regarding the eutrophication of rivers and streams.
34: “excessive”, why is it excessive? They are mostly unused and perhaps transported out of reach of crops. Perhaps “unused” is a better word here? This reviewer has a different interpretation of “excessive” versus “unused.” Saying unused implies there may be some management action such as fertilizer timing or cover crops. Excessive sounds like we just put too much down without consideration.
39: Agreed! This dataset is a nice contribution. Thanks.
53: “over 1000” is confusing. Suggest to explicitly state actual number as you do in abstract.
113: is it 12%? This reviewer considers that significant, not small.
230: Maavara did not focus on rivers, 12% trapped by reservoirs. Citation does not support statement.
270: What is a “hydrologic station”? Do you mean a streamflow gaging station? Discrete P data paired with streamflow, this reviewer would call that a “water-quality station.” Is it streamflow, WQ, or both? “Hydrologic” suggests no WQ data. This reviewer suggests picking consistent terminology.
290: Agreed!
293: What about trends? What is the period of record? If 2015-2020, this is for more recent understanding, so trends are not as priority.
309: That’s a good number to have. Nice work. But what time period does it represent?
Citation: https://doi.org/10.5194/essd-2025-743-RC2
Data sets
Riverine Phosphorus Gain and Loss Yiming Wang, Xuesong Zhang, Kaiguang Zhao, Robert D. Sabo, Yuxin Miao, and Christopher M. Clark https://doi.org/10.6084/m9.figshare.28509317
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 212 | 95 | 20 | 327 | 55 | 23 | 25 |
- HTML: 212
- PDF: 95
- XML: 20
- Total: 327
- Supplement: 55
- BibTeX: 23
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents a comprehensive dataset of riverine phosphorus (PO43- and total phosphorus, TP) gain and loss across catchments in the conterminous United States (CONUS). By integrating water quality observations, streamflow data, and hydrological connectivity (NHDPlus), the authors derive spatially explicit estimates of phosphorus loads, gains/losses, and source contributions at the HUC12 scale. Overall, the manuscript is well written, methodologically sound, and highly relevant to the ESSD community. The dataset fills an important gap in large-scale characterization of riverine phosphorus dynamics and will be valuable for watershed modeling, nutrient management, and environmental assessment. The methods are generally robust, and the data product is clearly described and accessible.
I have several specific comments that I hope the authors will address before acceptance for publication.
1. Clarification of the use of the LOADEST model
The use of LOADEST is appropriate, but several points would benefit from clarification: The reported r² values (0.76 for PO₄³⁻ and 0.83 for TP) are reasonable, but are there spatial patterns in model performance? Would it be useful to include a distribution of r² in the SI?
2. Refine equation 2.
The estimation of nonpoint source TP (Eq. 2) is a key contribution. The manuscript states that this is a “lower-end estimate” due to ignoring in-stream removal. Consider explicitly rewriting Eq. (2) with all assumptions clearly stated.
3. Explain the aggregation/disaggregation of data at different scales (HUC12 vs HUC8 vs HUC4)
There are multiple spatial scales used, including Gain/loss at HUC12 groups, point sources at HUC12s, NIP inputs at HUC8s, agricultural inputs at HUC4s. Please clarify how aggregation/disaggregation was handled when combining datasets across scales.
4. Spatial Coverage and Representativeness
The datasets cover ~4.9 million km² for PO₄³⁻ and ~6.1 million km² for TP, representing approximately 61% and 76% of the CONUS area, respectively. The western U.S. is notably underrepresented. While this is acknowledged, the authors should provide a more explicit spatial characterization of data gaps, including a figure showing the density of HUC groups or the proportion of area covered per HUC2 region. This would help users understand where inferences are most reliable.
5. Editorial Issues
The abstract states "51,394 PO₄³⁻ and 285,675 TP concentration data points" — consider rephrasing to "concentration measurements" for clarity.
In Section 2.2, longitude/latitude ranges for CONUS appear reversed.