the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Jun 2025
Warmer growing seasons improve cereal yields in Northern Europe only with increasing precipitation
Abstract. Crop yields depend on climatic conditions such as precipitation and temperature and their timing before and during the growing season. At high latitudes, climate change could lengthen the growing season and provide more suitable temperatures, but also expose crops to more frequent damaging conditions. We quantified the response of regionally-averaged 1965–2020 winter and spring cereal yields in Sweden to a wide set of descriptors of climatic conditions. With statistical models, we explored the role of both short-term and average conditions over physiologically relevant developmental stages, as well as of a proxy of water availability during the period prior to the main growing season. Temperature and precipitation or dry spell lengths for the entire growing season explained 75–85 % of yield variability, performing better than short-term potentially damaging conditions. Low precipitation or extended dry spells combined with high temperatures and, conversely, high precipitation sums with cool temperatures decreased yields for all crops. Our findings suggest that under climate change crop yields will be reduced in Sweden, unless warming is accompanied by increase in precipitation during the main growing season. With unaltered or reduced growing season precipitation, benefiting from warmer temperatures caused by climate change will require adaptation measures.
- Preprint
(1542 KB) - Metadata XML
-
Supplement
(1718 KB) - BibTeX
- EndNote
Status: open (until 30 Nov 2025)
-
RC1: 'Comment on egusphere-2025-1982', Anonymous Referee #1, 16 Aug 2025
reply
-
AC1: 'Reply on RC1', Faranak Tootoonchi, 19 Sep 2025
reply
Responses can be found in the attached pdf.
-
AC1: 'Reply on RC1', Faranak Tootoonchi, 19 Sep 2025
reply
-
CC1: 'Comment on egusphere-2025-1982', Martin Skoglund, 24 Oct 2025
reply
General comments
This study addresses a highly relevant and timely topic, and I find the authors’ attempt to integrate short-, medium-, and long-term climatic influences on crop yields both commendable and innovative. The analysis represents a significant step forward compared to previous work by systematically evaluating multiple climatic indicators across different time-scales and their interactions across different crop types and periods.
However, I have several concerns regarding the model specification, which may substantially affect the results and their interpretation. These issues primarily concern how the models account for spatial heterogeneity in climatic responses across Sweden. In addition to this, I have some other smaller concerns regarding the interpretation of the results when using relative yield indicators as well as the formulations of the model equations.
- Model specification
My main concern is that the models do not properly account for the well-known spatial differences in the temperature–precipitation dependency of crop yields between northern (N) and central/southern (C/S) Sweden. These regional contrasts were already formally identified by Wallén (1917, 1918) and have been confirmed in subsequent studies dealing with both historical and recent periods (e.g., Edvinsson et al. 2009; Skoglund 2022, 2023; Sjulgård et al. 2023).
In short:
- In northern Sweden (Norrland and Dalarna), higher summer temperatures are positively associated with yields, whereas increased precipitation often has a negative, albeit small, effect.
- In central/southern Sweden (counties south of Dalarna/Gävleborg), the relationship is generally the opposite: higher temperatures tend to reduce yields, while greater precipitation tends to increase them.
These contrasting relationships imply that aggregating N and C/S counties into a single model (this article) or time-series (see Holopainen et al. 2012), without explicitly accounting for these systematic differences, introduces biases and will underestimate the true climatic sensitivity of year-to-year yield variability.
This issue is especially relevant for spring-sown crops, such as barley, which are cultivated across the entire country, but also potentially for winter crops and wheat to a lesser extent (see Sjulgård et al. 2023). For winter crops and wheat, the shift from positive (negative) or negative (positive) relationships with temperature (precipitation) tend to occur only in the southernmost counties, if at all.
In the Discussion, the authors write:
“The explanatory power of climatic conditions were lower for oats and spring barley yields compared with spring and winter wheat … A possible explanation is that wheat yield data refer to southern Sweden only, whereas spring barley and oats are grown under a wider range of latitudes and hence climatic conditions …”
Here the authors themselves imply that there is a possible aggregation bias, that lowers the explanatory power for oats and barley. The issue is less relevant for (spring and winter) wheat that is mainly grown in C./S. Sweden with a more homogenous climatic signal (see also the results for barley and wheat yields in Holopainen et al. 2012 where the same type of aggregation error is made). In an analysis where relationships are estimated at the county-level, where the systematic difference between N. and C./S. is largely accounted for (except perhaps when considering border counties such as Dalarna/Värmland/Gävleborg), it can clearly be seen that year-to-year climatic fluctuations have a much greater explanatory power in N. Sweden compared to C./S. Sweden (Sjulgård et al. 2023; Skoglund, 2022; 2023). In your model, this is mainly introduced as a random effect, which brings the oat and barley models to similar levels of explanatory power as the wheat models. However, because the random effects model assumes that group-level differences are uncorrelated with the explanatory variables, this treatment is inappropriate when the between-county differences are themselves driven by climate–yield dependencies.
Possible alternatives that address the aggregation bias:
- Including an interaction between region (N, C, and S or N and C/S) and key climatic variables.
- Fit separate models for N, C and S or N and C/S.
- A random-slope model that allows the effects of temperature and precipitation to vary by county.
Relevance of results: Another issue, that is related to the model specification but only becomes an issue in regards to the interpretation of the results is that since you are treating yield as a relative variable (i.e., tonnes per hectare), instead of absolute production levels, the lumping together of N. and C./S. Sweden also obfuscates the social relevance of the results as the overwhelming majority of grain production occurs in C./S. Sweden.
Equations: As a previous reviewer mentioned, in your model equations (eq. 1–3), you describe yield as, Y, when it should in fact be indexed as time- and location-variant, Yit. Furthermore, all explanatory variables should also include it since they describe a given variable at time t and location i. I do not believe the choice to make it only Y makes it clearer as the authors suggests as we now are several reviewers who had the same objection.
In summary, this paper makes a valuable contribution and offers promising methodological advances, but the current model specification overlooks a key structural feature of Swedish agroclimatic variability. Addressing these well-known north–south contrasts in yield–climate relationships is essential to obtain unbiased estimates and to strengthen both the statistical validity and applied relevance of the findings.
References
Holopainen, J., Rickard, I. J., & Helama, S. (2012). Climatic signatures in crops and grain prices in 19th-century Sweden. The Holocene, 22, 939–945.
Sjulgård, H., Keller, T., Garland, G., & Colombi, T. (2023). Relationships between weather and yield anomalies vary with crop type and latitude in Sweden. Agricultural Systems, 211, 103757.
Skoglund, M. K. (2022). Climate variability and grain production in Scania, 1702–1911. Climate of the Past, 18, 405–433.
Skoglund, M. K. (2023b). Farming at the margin: Climatic impacts on harvest yields and agricultural practices in Central Scandinavia, c. 1560–1920. Agricultural History Review, 71, 203–233.
Wallén, A. (1917). Sur la corrélation entre les récoltes et les variations de la température et de l’eau tombée en Suéde. Kungl. Svenska vetenskapsakademiens handlingar, 57, 1–87.
Wallén, A. (1918). Sambandet mellan klimat och skörd i Sverige. Ymer, 1, 1–23.
Citation: https://doi.org/10.5194/egusphere-2025-1982-CC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 706 | 94 | 24 | 824 | 34 | 16 | 32 |
- HTML: 706
- PDF: 94
- XML: 24
- Total: 824
- Supplement: 34
- BibTeX: 16
- EndNote: 32
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript systematically evaluated the effects of different climate factors (temperature, precipitation) on crop yield before and during the growing season, focusing on winter and spring cereal yields in Sweden from 1965 to 2020. Using county-level yield data and a range of physiologically relevant climate indictors for growth stages, the study found that warmer temperatures only benefit yields if accompanied by increased growing-season precipitation. This study, aligned with the focus of Biogeosciences on climate-ecosystem interactions, and provides recommendations for yield management in the context of climate warming in Sweden's high latitudes. However, the manuscript requires further improvement in its variable selection strategy, model specification, and reproducibility of the results. The article's structure and language should also be refined.
Specific Comments:
Technical Corrections: