Supplementary Material
Study Area
The 33 counties considered are located in the Pampas region of Argentina. The Pampas of Argentina (located between 28 and 40◦S and 68 and 57◦W) extends over approximately 34 million hectares (Mha) encompassing the province of Buenos Aires and parts of the provinces of Entre Rios, Santa Fe, Cordoba and La Pampa and ca. 250 counties (Fig. S1). The area includes five phytogeographic districts of the Rio de la Plata grasslands, which covers 70 Mha of Argentina, Brazil and Uruguay (Ghersa and Leon, 1999). Mean annual precipitation (P) ranges from 400mm in the southwest to more than 1200mm at the northeast. The rainfall regime tends to be monsoonal at the northwest and evenly distributed in the southeast (Hall et al., 1992). Mean annual temperature displays a north–south gradient of approximately 5 ◦C, from 13.5 to 18.5 ◦C. Mollisol is the dominant soil order over the area (SAGPyA, 1990) and Argiudols and Haplustols are the most representative soils of the region. Topography is markedly flat with the exception of some subregions with slight ondulations (e.g. Rolling Pampas).
Agriculture in the Pampas started around 1880 when immigrant farmers started renting land devoted to cattle or sheep production and began sowing wheat for 3-4 yr periods at the end of which Lucerne was sowed (Barsky and Gelman, 2001). Since the diffusion of no-tillage systems and genetically modified crops in the 1990’s, and the more recent increased in soybean international prices, an intensification of agriculture took place. This intensification was characterized by the replacement of the mixed cattle and crop systems by continuous cropping. According to the archives of the Ministerio de Agricultura Ganaderia y Pesca from Argentina (SIIA) soy, maize and wheat are the most important crops. Soy in the Pampas can be sown early or late in the season (in the latter late soy generally follows a wheat crop). For an area of ca. 3Mha from the Rolling Pampa, the Flooding Pampa and the Inland Pampa, Caride et al. (2012) found that 5-years rotations were dominated by soy, wheat-soy, maize (19%) , followed by soy monoculture (14%), soy, wheat-soy (7%), soy, maize, soy, wheat-soy (6%) and soy, maize (6%). Grasslands occupied one third of the studied area and crop-pasture rotations accounted for 16%.
In general, extensive crops are responsive to P, T and DTR, given that P is a proxy of plant water availability, T mainly controls development and respiration and DTR accounts for different processes’ sensitivities to Tmax or Tmin and because DTR is positively associated to solar radiation (Bristow and Campbell 1984).
Based on local knowledge and previous studies (Hall et al. 1992) we assumed a growing season from June 1st to December 15th for wheat, September 15th to March 15th for maize and November 1st to April 15th for early soy. Although soy is planted in two different seasons we considered all soy as early season because national statistics do not discriminate between early and late soy. Considering all soy as early soy certainly introduces some error whose magnitude is constrained by the overlap between both growing seasons (coarsely there is 1,5 month difference at sowing while harvests dates tend to converge) and the amount of late soy sowing (that is roughly equal to wheat area, on average 5 Mha –i.e. between 10 and 20% of the total area sown with soy-).
Table S1: Median estimates of the impact of P, T and DTR trends from 1971 to 2012 on maize wheat and soy yields with the climate dataset considering one month previous to sowing.
Crop / P / T / DTR / T & P / T, P & DTRMaize / -0.34 / -4.60 / 0.01 / -4.89 / -4.90
Wheat / -0.30 / -2.85 / -2.27 / -3.13 / -5.34
Soy / -0.03 / -2.58 / 0.18 / -2.60 / -2.46
Figure S1. Spatial distribution of the counties considered (shown in grey). Capital letters at the beginning of the names refer to provinces: Buenos Aires (BA), Cordoba (C), Entre Rios (ER), La Pampa (LP) and Santa Fe (SF).
Figure S2. Wheat, maize and soy yield trends from 1971 to 2012. Values are expressed in kg.ha-1.y-1 (upper panel) and %.ha-1.y-1 (lower panel).
Figure S3. Observed wheat maize and soy growing season P, T and DTR trends between 1971 and 2012. Values correspond to the ratio of the 42-year trend devided by the standard deviation for the same period. Trends are only shown for counties that have available long term daily weather measurement station data.
Figure S4. Relationship between yield impacts estimated by local first difference models and yield impacts estimated by panel model.
Figure S5. Sensitivity of maize growing season P, T and DTR trends to the starting and ending years. Each line corresponds to one county.
Figure S6. Relationship between yield temporal trends calculated using different thresholds for harvested area (1000 ha vs 10000 ha) for the 3 crops (n=77).
References
Barsky, O, Gelman, J (2001) Historia del Agro Argentino. Grijalbo Mondador, Buenos Aires
Caride, C, Piñeiro G, Paruelo JM (2012) How does agricultural management modify ecosystem services in the argentine Pampas? The effects on soil C dynamics. Agriculture, Ecosystems & Environment 154:23-33
Ghersa, CM, Leon RJC (1999) Successional changes in agroecosystems of the Rolling Pampa. In: Walker, LR (ed) Ecosystems of Disturbed Ground. Elsevier, Amsterdam, pp. 487–502
Hall, AJ, Rebella CM, Ghersa CM, Culot J (1992) Field-crop system of the Pampas. In: Pearson CJ, (ed) Field crop ecosystems. Elsevier, Amsterdam, pp 413–450
SAGPyA, (1990) Atlas de Suelos de la República Argentina. Secretaría de Agricultura Ganadería Pesca y Alimentación, Buenos Aires.
9