Biological parameters for germination of selected summer weed species: first step to transfer the hydrothermal model AlertInf to Croatia

The efficacy of weed management depends on the correct control timing according to the seedling emergence dynamics. Since soil temperature and soil moisture are two main factors that determine weed germination, the hydrothermal time model can be used to predict their emergence. The aim of this study was to estimate the base temperature (Tb) and base water potential (Ψb) for germination of Chenopodium album, Amaranthus retroflexus, Setaria pumila and Panicum capillare collected from fields in continental Croatia and then to compare these values with those of Italian populations embedded in the AlertInf model. Germination tests were performed at seven constant temperatures (ranging from 4 to 27°C) and eight water potentials (0.00 to 1.00 MPa). Estimated Tb and Ψb were 3.4°C, -1.38 MPa for C. album, 13.9°C, -0.36 MPa for A. retroflexus, 6.6°C, -0.71 MPa for S. pumila and 11.0°C, -0.87 MPa for P. capillare, respectively. According to the criterion of overlap of the 95% confidence intervals, only Tb of C. album, and Ψb of A. retroflexus were similar between Croatian and Italian populations. Further field experiments should be conducted in the Croatian field to monitor weed emergence patterns of C. album and to calibrate the AlerInf equation parameters.


Introduction
Integrated Weed Management recommends the use of post-emergence herbicides when it is possible to select an effective herbicide or combination of herbicides based on the weed flora composition. Although such an application can be tailored to the specific botanical composition of the real weed flora, efficient control is highly dependent on timing according to weed emergence dynamics [1]. Therefore, knowledge of the timing and duration of weed emergence could help to achieve effective herbicide application without subsequent corrective treatments [2]. In addition, compared to standard management practice, it allows for lower herbicide application and lower weed control costs [3]. Weed emergence data are a basis for the development of predictive weed emergence models. These models provide the percentage of cumulative weed emergence achieved daily in the field, with the aim of suggesting the best time for farmers to control weeds [1]. Several predictive weed emergence models have been developed and are currently available for growers of maize [4,5], soybeans [6,2] and winter cereals [7,8] in Europe and the United States. These models are often based on the concept of thermal time (TT) or hydrothermal time (HTT) [9], depending on whether they consider only temperature (TT) or temperature and soil moisture (HTT) as triggers for germination. HTT models taken from Masin et al. [6] and for location Zagreb from Croatian Meteorological and Hydrological Service. Average annual precipitation in Zagreb is 861.1 mm with minimum precipitation in February (44.6 mm) and maximum in September (101.6 mm). Average annual temperature is 11.8°C, with minimum temperature in January (-3.2°C) and maximum in August (25.0°C). Padova has an average annual precipitation about 850 mm uniformly distributed throughout the year. Average annual temperature is 12.2°C, with temperature increases from January (average minimum value: -1.5°C) to July (average maximum value: 27.2°C)

Seed material
The seeds of S. pumila, P. capillare, A. retroflexus and C. album were hand-picked from plants in maize fields at physiological maturity. The seeds of C. album and S. pumila were collected at the Experimental Station of the University of Zagreb Faculty of Agriculture, Sasinovecki Lug (45°50'59.6"N 16°09'53.9"E), while the seeds of A. retroflexus were collected at the Experimental Station Maksimir (45°49'34.3"N 16°01'49.8"E) and P. capillare were collected at the site Lipovec Lonjski (45°44'51.9"N;16°23'12.4"E). The collected seeds were brought to the laboratory, cleaned, sieved and stored in paper bags in the refrigerator (4°C) until the beginning of the experiment.

Germination experiments
Experiments to estimate base temperature and base water potential for germination were conducted at University of Padova, Department of Agronomy, Food, Natural resources, Animals and Environment and University of Zagreb, Faculty of Agriculture, Department of Weed Science from 2013 to 2020. Prior to the start of the experiments, a preliminary germination test in the climate chamber (W87R, KW Apparecchi Scientifici SRL, Italy) at constant temperature (25°C) and photoperiod 12 h:12h (day: night) was conducted to check the germination capacity of the seeds. Seed populations that achieved a germination higher than 60% were included in further studies.
The estimation of the base temperature of four weed species was performed by testing the germination at six or seven constant temperatures with photoperiod 12h:12h (day: night) simultaneously in the different climatic chambers. To prevent the growth of pathogens on the seeds and on filter paper, seeds were sterilized with 1% hydrogen peroxide and washed with distilled water. Three replicates of 100 seeds or five replicates of 50 seeds were placed in the Petri dish on Whatman® filter paper covered with 5 ml distilled water and sealed with parafilm. The initial temperature was defined for each weed species as one degree lower than the base temperature previously established from the literature [11, 18 -20]. Therefore, C. album and S. pumila were tested to a constant temperature of 4, 8, 12, 16, 20, 24, 28°C. Furthermore, the germination of A. retroflexus was tested at 9, 12, 15, 18, 21, 24, 27°C and P. capillare at 6, 9, 12, 15, 18, 21, 24, 27 and 30°C. To estimate base water potential of each species, germination test was carried out exposing seeds to different level of water potential, that is different level of water availability. Three replicates per 100 seeds or five replicates per 50 seeds were placed at eight different water potential solutions. For this purpose, polyethylene glycol (PEG) 6000 (Sigma-Aldrich Chemie GmbH 25322-68-3) was used to prepare solutions with eight water stress levels: 0.00 (pure distilled water), -0.05, -0.10, -0.25, -0.38, -0.50, -0.80, -1.00 MPa according to Michel and Kaufmann [21]. The seeds were placed in transparent plastic containers 10 cm in diameter and 7 cm high, as described by Masin et al. [1]. Containers with 50 ml prepared solution were placed at a constant temperature of 22°C and a photoperiod of 12h:12h (day: night).
In both germination experiments, seeds were defined as germinated when the seed radicle was 1 mm long. Germinated seeds were counted and removed twice daily for seeds incubated at temperatures above 20°C and all water potentials above -0.38 MPa and once daily for seeds temperatures below 20°C and water potential below -0.38 MPa. The germination test was considered complete when no germination was detected for 10 consecutive days. Germination test lasted from 9 to 95 days depending on the temperature or water potential and tested weed species.
The temperature in the climate chambers was recorded hourly with temperature data loggers (HOBO UA-001-08, Onset Computer Corporation, Bourne, MA). Temperature deviations ± 0.5°C were considered acceptable.

Statistical Analysis and Statistical Methods
Mean percentage of germination was calculated for each treatment. The germination dynamics curve was generated using the logistic function in the Bioassay97 statistical program [22] to determine initial (t10), medium (t50) and final (t90) germination time using the formula: CG = 100/ 1 + exp {a [ ln (t+0.0000001)-ln (b)]} Where CG is the percentage of cumulative germination, t time expressed in days, a is the slope of the curve and b is the inflection point. The initial (t10), medium (t50) and final (t90) germination time i.e., the time it takes 10, 50 and 90 % of the germinating seeds to germinate, are determined by the slope of curve (b). The effect of temperature and water potential on germination percentages and germination dynamics (t10, t50 and t90) was analyzed by means of variance analysis (ANOVA). After the significant F-test, the LSD test for P = 0.05 was used to compare the mean values.
The biological germination parameters were determined using germination dynamics data at different temperatures and water potentials for each species studied. The reciprocal of t50 (1 / t50) was used to establish the linear regression line against the incubation temperature or water potential [1]. The values of Tb and Ψb were presented as the point where the linear regression line intersects the abscissa. The 95% confidence intervals for Tb and Ψb were determined using the bootstrap method [23]. The values obtained for the biological parameters of the Croatian populations were compared with the values of Italian built into the AlertInf model, according to the criterion of overlap of the 95% confidence intervals [1]. If there is no overlapping of the confidence intervals between the two populations, a significant difference is determined.

Weeds germination at different temperatures and water potentials 2
Analysis of variance showed significant influence (P< 0.0001) of studied temperatures and water potential to weed germination (Table 1 and  3  2). 4 The germination of the species studied varied between 0.67 and 99% depending on 14 incubation temperature or weed species. Among the weed species tested, C. album 15 showed the ability to germinate at the lowest temperature (4°C), while other species 16 started to germinate at 8°C (S. pumila) or 12°C (A. retroflexus and P. capillare). The 17 highest germination percentage of all four species was reached at 24°C: C. album (98%), 18 A. retroflexus (95%), S. pumila (93%) and P. capillare (93%). However, A. retroflexus 19 achieved similar germination percentage also at 27°C, or C. album and S. pumila at 20 temperature range 16 -24°C. Germination began to decrease as the temperature dropped, 21 but this process was species-specific. Germination of C. album and S. pumila decreased 22 at temperatures ≤ 8°C. Germination percentage of P. capillare decreased at temperatures 23 ≤ 15°C and then plunged at 12°C, while A. retroflexus showed a sharp reduction of ger-24 mination even at 15°C. These results indicate that incubation temperatures affect greatly 25 the germination of tested weed species.
26 Similarly, water potential greatly affected the germination of weed species. For all 27 tested weed species germination began to decrease as the water potential dropped (Table  28 2). Germination of S. pumila and C. album showed a reduction from water potential of 29 -0.50 and -0.38 MPa, respectively. A. retroflexus presented a first decrease of germination 30 at -0.10 MPa and then a strong inhibition from -0.25 MPa. P. capillare ceased germination 31 at -1.00 MPa. All weed species presented almost no germinated seeds at -0.80 and -1.00 32 MPa. Taking together, germination of all species decreased significantly at lower water 33 potentials, but the ability for germination at different water potential was also species 34 specific. 35

Germination dynamic in response to different temperatures and estimation of base temperature 38
Daily recorded germination data was used to obtain germination dynamic curves at 39 each studied temperature. Estimated time required for t10, t50 and t90 is expressed in days 40 (d) as decimal number (Table 3). The germination dynamics were influenced by temper-41 ature in all tested species. A decrease in temperature led to an increase in the number of 42 days required for the start and end of germination for all species. The duration of ger-43 mination varied between 0.6 and 76.0 days depending on the incubation temperature and 44 the species studied. At 24°C the initial germination (t10) was shortest for A. retroflexus 45 (1.0 d) and longest for C. album (3.3 d). At the same temperature, A. retroflexus needed 46 1.4 d to achieve medium germination (t50), while C. album continued with a prolonged 47 trend with the longest t50 value of 4.5 days. In contrast, no statistical difference was found 48 between two monocotyledonous species P. capillare and S. pumila in the medium ger-49 mination (t50) at a temperature of 24°C. The end of germination (t90) was reached for A. 50 retroflexus in 1.8 d, while C. album and P. capillare finished germination in 6.2 and 7.0 d. 51 As expected, lower temperatures prolonged the germination of all investigated spe-52 cies. Due to the low germination capacity of A. retroflexus at 12°C (0.67%) it was not 53 possible to establish a germination curve as was the case at other temperatures. The ini-54 tial, medium and final germination of the other three species at 12°C varied between 5.83 55 and 24.38 days. C. album and S. pumila started germination at 5.83 and 8.98 d, while P. 56 capillare extended the start of germination to 12.17 d. S. pumila was the species that 57 reached before the others the end of germination (t90 at 12.3 d), while P. capillare set the 58 end of germination to 24.38 d. 59

Germination dynamic in response to different water potential and estimation of base water 67 potential for tested weed species 68
The duration of germination of all weeds varied between 0.4 and 32.3 d, depending 69 on the incubation water potentials and the species tested. In general, the duration of 70 germination increased with the decrease in water potential. The germination was ex-71 tended in the range depending on the species (Table 4). 72 A. retroflexus showed the highest sensitivity to water stress. After a very low ger-73 mination at a water potential < -0.25 MPa, it was even not possible to estimate the ger-74 mination dynamic curve. At a water potential > -0.25 MPa, germination lasted from 0.5 to 75 3.6 d (t10-t90) and at -0.25 MPa, germination lasted 18.6 d (t90). Other species required 76 longer time to reach initial germination phase (t10) at a water potential < -0.25 MPa, but 77 then they were able to maintain similar germination dynamics until -0.38 and -0.50 MPa 78 for S. pumila, C. album and P. capillare, respectively C. album was the only species with 79 the ability to germinate at all investigated water potentials. 80  It is important to underline that for P. capillare it was not possible to use the logistic regression model to identify the t50 at -0.80 MPa, due to the low germination (Table 2). However, a value of 1/t50 close to zero was used at -0.80 MPa to estimate the base water potential. It was necessary to avoid underestimation of the base parameter.

Comparison of biological parameters of Italian and Croatian populations 88
According to the criterion of overlap of the 95% confidence interval [1] between 89 Italian and Croatian populations, two out of three species tested have similar values in an 90 estimated parameter (Table 5). For P. capillare it was not possible to make a comparison 91 because the biological parameters of the Italian population of this species have not yet 92 been estimated. 93 The Croatian population of A. retroflexus had a 1.6°C higher base temperature 94 compared to Italian populations, and the overlap was not found even if the extreme of 95 the two confidence intervals were close. So, these two values of Tb can be considered as 96 statistically different. In the Croatian population of C. album Tb was 0.8°C higher than in 97 the Italian population, but the confidence intervals overlapped. These two values are 98 therefore not statistically different. In contrast, Tb value estimated for the Croatian pop-99 ulation of S. pumila is 3.81°C lower compared to the Italian population and it was found 100 that they differed significantly. 101 Base water potential of A. retroflexus was 0.05 MPa higher for the Croatian popula-102 tion compared to Italy but no significant difference was found. Lower base water poten-103 tial was determined for the Croatian population of C. album, and higher for S. pumila, 104 anyhow significant differences from the Italian population were found in both cases.
105 Taken together, similar values between Italian and Croatian populations were found 106 only for C. album regarding Tb and for A. retroflexus regarding Ψb. 107

Discussion
In the present study biological parameters for germination (Tb and Ψb) of four summer weed species collected in Croatia were estimated. Germination tests at different temperatures showed a species-specific preference for higher or lower temperatures. Species are ranged from less to more thermophilic as follow: C. album > S. pumila > P. capillare > A. retroflexus. This is consistent with previous studies in which A. retroflexus also germinated best at temperatures > 25°C [24] and S. pumila at temperatures of 24.5 to 34.9°C [25]. In the present study P. capillare had the highest germination at temperatures of 18-24°C again consistent with previous study where P. dichotomiflorum achieved the highest germination capacity at temperatures of 25°C [26][27][28][29] or P. miliaceaeum at temperatures between 18 and 25°C [30,31]. Optimal temperature for germination of C. album has been reported between 15 and 25 °C [32], which is again in line with the germination data of the present study (Table 1).
If we related the temperature preferences that define germination to the time of emergence in the field, as suggested in a previous study [2], the species from the present study could be divided into three categories: early (C. album), middle (S. pumila and P. capillare) and late emerging species (A. retroflexus). The germination dynamic data shown in this study (Table 3) also reflected the species-specific sensitivity to different temperatures. In particular, A. retroflexus had the shortest germination at all temperatures investigated. For example, the mean germination of this species stopped completely after 5.7 days, which is slightly faster compared to other species in the study (Table 3). Germination stopped completely at 12°C confirming its thermophilic behavior [2]. Tb values for A. retroflexus is estimated to be 13.9°C, which is the highest Tb among the species tested in the present study. This is the reason why later emergence is observed for A. retroflexus in maize fields compared to other weeds studied [33]. This Tb is slightly higher than the values around 12°C reported for Italian and Iranian populations [1,11] and even lower values, i.e. 10.5 and 8.9°C, have been reported for other populations from Germany and France respectively [19,34]. In contrast, our study estimated that Tb for C. album is 3.4°C making it the species with the lowest temperature requirement confirming its early germination behavior [2]. This value is similar to the Tb reported for Italian and Dutch populations [1, 35] while a significantly higher value was described for a French population [19]. The value of Tb (6.6°C) estimated for the Croatian population of S. pumila is lower than the range of values (8.6-10.4 °C) reported in previous studies on populations from Italy, French and California [1, 19,30]. In our study we also estimated Tb for P. capillare, and as far as we know, this is the first report globally on base temperature for this weed species. We found that 11°C is the base temperature for P. capillare. Our results are consistent with the study where the minimum temperature for germination of P. miliaceaum was estimated to be also 11°C [31].
Since temperature is not the only factor triggering germination in the present study, we also observed the germination capacity of the same weed species under different water potentials. Depending on water requirements, the species in this study are ranged from less to more tolerant as follow: A. retroflexus (Ψb -0.36 MPa) < S. pumila (Ψb -0.71 MPa) < P. capillare (Ψb -0.86 MPa) < C. album (Ψb -1.38 MPa) (Figure 2). In addition to the study performed to calibrate AlertInf model [1], only a single study conducted in France has already determined Ψb for these species [19]. The value of Ψb estimated for the Croatian population of A. retroflexus is similar to the Ψb used in AlertInf for a population from Northern Italy, while lower values were reported for other populations from Central Italy (-0.62 MPa) or France (-0.95 MPa) [1,19]. Regarding Ψb for S. pumila, an almost identical value was described for a French population (-0.75 vs -0.71 MPa) while a lower value was determined for the Italian population included in AlertInf [1,19]. Finally, the Croatian population of C. album had a lower Ψb in comparison with the values previously reported for both Italian (-0.96 and -1.04 MPa) and French (-0.80 MPa) populations [1,19]. For the species P. capillare as far as we know there are no data of Ψb in the literature. Germination behavior of the investigated species at different temperature and water po-tential regimes shows that the species with a better tolerance to lower temperatures also had a better tolerance to lower-water potential.
The main objective of the present study was to compare the estimated values of the Croatian population with Italian populations of C. album, S. pumila and A. retroflexus included in the hydrothermal model AlertInf. We wanted to examine if the Tb and Ψb values estimated in this study would be comparable with those in Alertinf [4] as a first step of transferring weed predictive model AlertInf in Croatian maize fields. Two out of three Croatian species had a parameter overlapping with the Italian population: C. album had similar Tb, but different Ψb, A. retroflexus different Tb but similar Ψb. and S. pumila differed in both germination parameters (Table 5).
Present study therefore showed that similar germination of Croatian and Italian population may only be expected for C. album since the Tb value was similar. However, this is valid only in conditions were soil water is not limited since Ψb differed significantly between two populations. Next step therefore will be to evaluate the weed emergence patterns of C. album in irrigated maize fields in Croatia and then try to calibrate the AlerInf equation parameters. Prediction of A. retroflexus and S. pumila by AltertInf model with its original parameters is unfortunately impossible, even in irrigated maize crops, since the Tb values differ statistically with those Italian presents in the hydrothermal model. This variability in germination parameters has also been documented in previous studies and usually explained as adaptation process of weed species to local climate conditions [13,36]. The annual air temperature in Zagreb is on average lower than those in Padova (Table 1). Moreover, compering the temperatures in spring (March-June) and summer (July-October) in Zagreb (6.4 -19.4°C; 21.1 -11.0°C) and Padova (9.0 -22.0°C; 23.1 -14.0°C) during a thirty-year period it is evident that Zagreb has lower air temperatures. Therefore, we expected that populations in Zagreb and the surrounding area would have lower Tb values compared to those in Padova as suggested earlier [1,12]. This was accomplished for S. pumila and C. album, however this phenomenon was not found for A. retroflexus, where Tb is higher for the Croatian (colder climate) than for the Italian population (warmer climate). Unfortunately, the complexity of weed seed biology, especially in the period of seed ripening, can influence the germination behavior of seeds. The involvement of various factors that determine the characteristics of the seed (position on the mother plant, micro-environmental conditions, availability of nutrients, etc.) can cause the difference in dormancy and germination requirements [37,38]. An attempt to implement the model in another agro-ecological area was also made by these Bürger and Colbach [15] for the FlorSys model. The difference in base temperature for different species was also species-specific and unable to find the pattern connected to climate conditions. They have found 4.3 lower Tb for C. album and 4.0 ° C higher Tb E. crus-galli in Germany compared to France.

Conclusions
C. album, A. retroflexus, S. pumila and P. capillare are highly distributed weed species globally [39] and in Croatian maize fields [40]. Present study offers the ability to implement the predictive emergence model AlertInf only for C. album in non-irrigated field. However, the results are valuable since it provides estimated biological parameters of four species, which have never been estimated before in Croatia. This is the first and obligatory step towards developing/transferring a model to predict their emergence.
However, further field experiment is necessary and should be performed in two directions depending on the weed species. First, for C. album or A. therophrasti since their estimated Tb are overlapping with Italian populations [16], the AlertInf has to be further validated by comparing the emergence of species in maize fields with those predicted with AlertInf. Secondly, the model should be adapted for species S. pumila and A. retroflexus since biological parameters differed significantly. And the third, AlertInf should be upgraded for P. capillare whose biological parameters have now been estimated for the first time.