| 18 May 2026
Dissolved organic carbon and nutrient leaching in temperate alley-cropping agroforestry and open cropland
Abstract. Alley-cropping agroforestry with short-rotation coppice (AF) has been identified as a sustainable alternative to highly industrialised agricultural systems, enhancing ecosystem functions, including the regulation of water quality. However, there is limited knowledge regarding the leaching of dissolved organic carbon (DOC) and nutrients in AF and treeless open cropland (OC). In this study, we compared the leaching of mineral nitrogen, dissolved organic nitrogen (DON), DOC, potassium, and phosphorus in AF tree rows (AFtree) and AF crop rows (AFcrop) with OC across three sites with different soil textures in Germany over four years (2019–2022). The objectives were to (1) quantify the fluxes of DOC and nutrients in AF and OC systems and (2) identify the primary drivers of leaching in both systems. At each site, we sampled soil-pore water from AFtree, AFcrop (at 1 m, 7 m, and 24 m from the AFtree) and OC monthly using suction cup lysimeters installed at 60-cm depth and analysed for DOC and nutrient concentrations. AFtree was unfertilised whereas AFcrop and OC had the same conventional fertilisation rates. Drainage fluxes were estimated using the Expert-N water sub-model. Our results showed that, in loam Phaeozem and clay Cambisol soils with mature AF (established for at least 12 years with ≥ 5-year-old AFtree in the second rotation), AFtree had N leaching of 0–0.5 kg N ha-1 yr-1 whereas OC had 1–4 kg N ha-1 cropping period-1. Mature AF in the loam Phaeozem and clay Cambisol soils had two to 10 times lower NO3− leaching at 1 m than at 7 m and 24 m AFcrop. The effect of the AFtree on leaching was limited to ≤ 7 m AFcrop. Overall, AF (the area-weighted average of AFtree and AFcrop) did not differ from OC in DOC and nutrient leaching fluxes. An exception was observed in the sandy Arenosol soil during the first two years following AF establishment, when the overall AF exhibited higher NO₃– and K leaching fluxes than OC, attributed to the large liquid manure application combined with soil disturbance from tree establishment and young tree roots. Across sites, precipitation and sand content were the dominant drivers of DOC and nutrient leaching in AFtree whereas precipitation was the only factor influencing leaching in AFcrop and OC. These results reinforce the importance of synchronising fertiliser input with the crop's demand in AFcrop as excessive fertilisation can override the tree root interception of nutrients at greater distances from the AFtree.
Choe et al. interpret a valuable multi-year dataset of nutrient fluxes in temperate alley-cropping agroforestry and adjacent cropland across three German sites differing in soil texture, age of agroforestry establishment, and management history. They found that while in longer-established agroforestry systems, tree rows had lower nitrogen leaching than in the crop alleys and in open cropland, localized effects did not translate into overall lower area-weighted leaching from the agroforestry system as a whole. At a more recently established site, nitrate and potassium leaching were initially quite high but declined over the study period.
Overall, I do not think that the authors' interpretations of the data are necessarily wrong, but I do have concerns about the lack of discussion around the confounding of soil texture, site, tree age, crop rotation, and fertilization. Notably, the most recently established agroforestry system occurs on the site with sandy soil, while the older systems are present on loam and clay soils. It is therefore not possible to determine whether the observed differences are caused by tree maturity, soil texture, other site-level differences, or some combination of these factors.
I am also wary of some of the statistical analyses. The stepwise regression and claims about "dominant drivers" of leaching seem too strong given the study only includes three sites with confounded variables. The ratio-based analysis also requires clearer justification. Finally, several mechanisms are discussed as though they were demonstrated by the study, particularly root development and lateral nutrient movement, when they were not measured directly.
I think that the manuscript has the potential to be a useful contribution, but these issues should be addressed prior to publication.
General comments:
Confounding among variables:
As mentioned, the main limitation of the study is that the variables used to explain differences across sites are confounded. The youngest agroforestry system is located on the sandy Arenosol, while the older systems are located on the loam and clay soils. The sites also differ in crop rotation, fertilization history, precipitation, drainage, previous management, and duration of monitoring.
The decline in nitrate and potassium leaching at the young sandy site is consistent with increasing root development and recovery following establishment. However, it also occurred alongside changes in crop type, fertilization, hydrology, and time since disturbance. Similarly, the very low nitrogen leaching beneath the older tree rows may reflect tree uptake, lower drainage, the absence of fertilization, differences in soil properties, or a combination of these processes.
The manuscript needs to acknowledge this limitation much more clearly. The results show that the sites differed and that these differences were associated with soil texture and agroforestry age, but the study cannot separate the effects of these variables.
Stepwise regression and "dominant drivers":
I am not convinced that the stepwise regression supports the conclusion that sand content and precipitation were the dominant drivers of leaching.
Sand content is essentially a site-level variable, with only three independent site values. A strong relationship across site-by-crop-rotation means does not provide independent replication of a soil-texture effect. It may instead reflect any of the other differences among the three sites. Tree age was excluded because of its collinearity with sand content and precipitation, but this does not resolve the confounding. It shows that these effects cannot be separated using the available design.
The effective sample size also appears small relative to the number of potential predictors considered. Stepwise selection under these conditions may produce unstable models and overstate the importance of the selected variables.
I suggest either removing the stepwise regression or presenting it explicitly as an exploratory analysis.
Ratio analysis:
The agroforestry-to-cropland ratios may become unstable when the cropland flux is zero or close to zero. This appears to contribute to some of the very large ratios and confidence intervals in Figure 2.
The bootstrap procedure also needs to be described more clearly. In particular, it is unclear whether uncertainty in both the agroforestry values and the cropland denominator was included. Because multiple agroforestry ratios also share the same mean cropland denominator, the authors should explain how this dependence was handled in the bootstrap analysis.
Causal interpretations:
The manuscript sometimes presents plausible mechanisms as though they were directly demonstrated by the study.
For example, the decline in nitrate and potassium leaching over time at the sandy site is attributed to tree-root development, while high early losses beneath the tree rows are attributed partly to lateral nutrient transport from the fertilized crop rows. Both explanations are reasonable, but neither root interception nor lateral transport was measured directly.
The authors should distinguish more clearly between the observed results and the mechanisms proposed to explain them. Statements that root maturation or lateral nutrient movement caused the observed patterns should be changed to state that the patterns are consistent with these mechanisms.
This is also important because leaching fluxes were calculated by multiplying measured concentrations by modeled drainage. Low fluxes beneath tree rows could therefore result from lower concentrations, lower drainage, or both. The Discussion should avoid treating low leaching fluxes as direct evidence of nutrient uptake by tree roots unless the concentration and drainage results support that conclusion separately.
Specific comments:
Line 548 "on the contrary"
Line 704: "tree roots were higher at 1 m"... Please specify what was higher.
Line 709: the crops had high biomass C-content which was added by tree litter C input? Perhaps a word is missing or needs to be changed here.
Line 715-716: Is this sentence saying that further research is needed or that litter decomposition/DOC stabilization is needed for increased C storage?
Line 741-742: I would rephrase this sentence as it is currently grammatically a bit awkward. For example: “At the system level, AF did not reduce nutrient leaching fluxes relative to OC at any of the three sites.”
Table 3: Consider using NA instead of 0 for entries that are not applicable, e.g., tree leaf litter values in open cropland. A value of 0 implies that these were actually measured.
A minor suggestion: I found the use of OC as an acronym for open cropland a bit confusing while reading, because it typically refers to organic carbon in a soil context. You might consider changing this to help similarly challenged readers.