We studied cover crops in maize-based cropping systems in the agricultural plain areas of Northern Italy (Lombardia region), to: quantify the growth of winter cover crops, nitrogen uptake, weed control capacity, and nitrogen credits for maize; to test the possibility of sensing cover crop biomass and nitrogen (N) concentration using aerial images, and to estimate N concentration using near-infrared spectroscopy (NIRS); and to quantify cover crop cultivation costs. We set up six experiments to compare different cover crop species in rotation with maize. The experiments started in September 2017 and ended in August 2019.
Cover crop growth in autumn was rapid for white mustard, tillage radish and black oat (which accumulated 1.5–3.0 t DM/ha until November), while legume cover crops (Egyptian clover, hairy vetch and purple vetch) had a lower crop growth rate (reaching 0.5–1.5 t DM/ha). Therefore, non-legume cover crops controlled weeds better compared to legume cover crops. Nitrogen uptake in autumn was highest for white mustard, tillage radish and hairy vetch (77-125 kg N/ha). Rye and black oat were intermediate, while Egyptian clover and purple vetch were lowest (35 kg N/ha). Rye and hairy vetch survived winter, while white mustard was always destroyed by winter frosts. We did not observe relevant effects of cover crops on maize yield and N uptake.
In one experiment, we also established relationships between vegetation indices (NDVI, Normalised Difference Vegetation Index; and CIg, Chlorophyll Index) obtained with a multispectral digital camera carried by an Unmanned Aerial Vehicle and cover crop biomass and nitrogen concentration. Increasing biomass above about 1 and 2 t DM/ha for NDVI and CIg, respectively, did not correspond to increasing values of the indices, thus reducing their predictive capacity. A new index (calculated using measured crop height) improved the predictions of cover crop biomass substantially (R2 = 0.64–0.86). Cover crop nitrogen concentration was poorly predicted by the vegetation indices.
We finally scanned cover crops with two NIR instruments, and used the spectra for chemometric elaborations (Partial Least Squares regression). Predictions of N concentrations on spectra of fresh materials were rather poor, while dried and ground samples provided better results (R2 = 0.86 for the bench instrument, and 0.70 for the portable instrument). Locally Weighted Regression further improved the R2 (0.90 and 0.84, respectively).
Cultivation costs ranged between 112 €/ha for white mustard (a winter-killed species that does not require termination and has a low seed cost) and 208 €/ha for rye (winter-hardy, with more expensive seed).
L. Bechini1, M. Ben Hassine1, D. Ditto1, D. Cavalli1, M. Corti1, A Perego1, E. Potenza1, T. Tadiello1, G. Cabassi2, L. Borrelli2, D. Paolo2, N. Pricca2, L. Degano3, R. Fuccella3, F. Introzzi3, G. Lussignoli4, L. Michelon4, D. Della Torre4, A. Merli4, S. Toninelli4, M. Motti5, A. Vigoni6, D. Sacco7, P. Marino1
1 University of Milano, Italy.
2 CREA-ZA, Lodi, Italy.
3 Fondazione Morando Bolognini, Sant’Angelo Lodigiano (LO), Italy.
4 Condifesa Lombardia Nord-Est, Brescia, Italy.
5 Agricola Motti, Orzinuovi (BS), Italy.
6 Sport Turf Consulting, Milano, Italy.
7 University of Torino, Italy.
26 pages, 10 figures, 2 plates, 3 tables, 15 references