Naslov
Vodno-toplinsko modeliranje za prognozu nicanja korova u kukuruzu
(Predicting weed emergence in maize with hydrothermal modelling)

Autori
Valentina Šoštarčić

Vrsta, podvrsta i kategorija rada
Ocjenski radovi, doktorska disertacija

Fakultet
Agronomski fakultet

Mjesto
Zagreb

Datum
11.03

Godina
2022

Stranica
73

Mentor
Maja Šćepanović ; Roberta Masin

Ključne riječi
biološki minimum ; biološki vodni potencijal ; korovi ; prognoza nicanja ; integrirano suzbijanje
(base temperature ; base water potential ; intergrated weed management ; predictive weed emergence models ; weeds)

Sažetak
Maize is the most widespread crop in Croatia and weeds are the main production limiting factors. In modern maize production herbicide application is performed after weed and crop emergence. Since weed species can differ in the time and duration of emergence to achieve appropriately timed weed control, it is necessary to determine the period in the field when the largest population of weed species is expected. Weed emergence prediction models are being developed to predict peak periods of weeds so that farmers can determine the appropriate time to apply herbicides. Soil temperature and soil moisture are the two main factors affecting weed emergence under field conditions. Therefore, hydrothermal models can be used to predict weed emergence in agricultural crops. Hydrothermal models summarize thermal units subtracted from the value of base temperature (Tb) when the soil water potential is above the value of base water potential of the species (Ψb). AlertInf is an Italian hydrothermal model for weed emergence prediction in maize developed in the Veneto region. The possibility of validating this model to Croatian maize crops was tested during this doctoral research. Prior to validate the model, estimation of germination parameters (Tb and Ψb of each weed species) is required. In the doctoral research, germination parameters of economically important weeds in maize crop in Croatia (Amaranthus retroflexus, Chenopodium album, Echinochloa crus-galli, Abutilon theophrasti, Setaria pumila, Panicum capillare and Ambrosia artemisiifolia) were estimated. Naimely, if the biological parameters of native and foreign populations of the same species differ it is necessary to calibrate the model. The estimated Tb and Ψb of of the studied species are: Chenopodium album (3.4°C ; -1.38 MPa), Abutilon theophrasti (4.5°C ; -0.67 MPa), Setaria pumila (6.6°C ; -0.71 MPa), Echinochloa crus-galli (10.8°C ; -0.97 MPa), Panicum capillare (11.0°C ; -0.87 MPa), Amaranthus retroflexus (13.9°C ; -0.36 MPa). No statistical difference was found between the Croatian and Italian populations of Abutilon theophrasti for both germination parameters. Therefore, the AlertInf model can be validated in the Croatian maize field without calibration. No statistical difference was found between the Croatian and Italian populations of Echinichloa crus-galli and Amaranthus retroflexus in the base water potential parameters. However, a statistical difference was found in the base temperature parameters for these two species. Therefore, AlertInf should be calibrated for these two species and validated for the base temperature parameters. Statistical differences between Croatian and Italian populations of Setaria pumila were found for both parameters studied (Tb and Ψb). In order to use the AlertInf model in Croatian maize fields, the model should be calibrated and validated for both studied parameters. Since the AlertInf model does not consider the species Panicum capillare, it should be updated and validated for this weed species in maize fields in continental Croatia. The emergence of Echinochloa crus-galli in Croatian maize fields was successfully predicted with Alertinf including estimated germination parameters of the native population. The overall performance of the model was evaluated by the root mean square error (RMSE) and modeling efficiency (EF). The RMSE is 1.69 and 1.38 for 2019 and 2020, respectively. In addition, EF is 0.97 and 0.98 in 2019 and 2020, respectively. With the calibrated model AlertInf it is possible to predict the emergence of Echinochloa crus-galli in maize fields in continental Croatia. The results obtained in the PhD thesis also have a practical value for maize growers and fit well with the EU Directive 2009/128/ EC on the sustainable use of pesticides and the European Commission's Green Deal initiatives to reduce pesticide use in agriculture.

Izvorni jezik
Engleski

Znanstvena područja
Biologija, Poljoprivreda (agronomija)

POVEZANOST RADA