Chitu, E. Topor, E. Paltineanu, C. R. Dumitru, I. M.Sumedrea, D. Chitu, V. Ionita, A. D. Filipescu, L.

Research Institute for Fruit Growing Pitesti - Romania,Research Station for Fruit Growing Constanta - Romania,„Politehnica" University of Bucharest - Romania

28th International Horticultural Congress, Lisbon, Portugal, 2010

ObjectivesObjectives The purpose of this paper The purpose of this paper waswas to develop a to develop a

simulation model that estimates not only the simulation model that estimates not only the phenological dynamics until the end of bloomphenological dynamics until the end of bloominging, , but also to estimate the occurrence probability of but also to estimate the occurrence probability of climatic accidents caused by climatic accidents caused by the the late frosts in late frosts in apricotapricot orchards orchards fromfrom the South-Eastern part of the South-Eastern part of Romania Romania during 1985 – 2008 period,during 1985 – 2008 period, under the under the climatic change climatic change conditionsconditions

Due to the changes of continental-temperate climate of Due to the changes of continental-temperate climate of Romania, the late frosts affecting apricot trees by Romania, the late frosts affecting apricot trees by damaging flowers are happening more often damaging flowers are happening more often

MATERIAL AND METHODSMATERIAL AND METHODS

A database of A database of phenological observationsphenological observations concerning the concerning the

calyx red,calyx red, first bloomfirst bloom and the end of and the end of petal fall petal fall (in the shuck), (in the shuck),

in in apricotapricot (‘ (‘Prunus armeniacaPrunus armeniaca L.’), ‘Umberto’, ‘C.R. 2-63’, L.’), ‘Umberto’, ‘C.R. 2-63’,

‘Sirena’, ‘Neptun’ and ‘Selena’ cultivars orchards‘Sirena’, ‘Neptun’ and ‘Selena’ cultivars orchards,, for a 24 for a 24

year period (1985-2008)year period (1985-2008),, was used. The orchards was used. The orchards

investigated were located within investigated were located within plainplain region of region of Constanta Constanta

(Valu lui Traian)(Valu lui Traian), , SSouthouth-Eastern-Eastern Romania, with trees of Romania, with trees of

various ages. various ages.

Weather dataWeather data: : Mean, maximum and minimum daily Mean, maximum and minimum daily

temperatures ftemperatures fromrom the February 1 the February 1stst to to the the May 31May 31stst,, during during

2424-year period -year period interval interval were were taken into considerationtaken into consideration..

The critical temperatures The critical temperatures for thefor the t threehree phenophases, according to the phenophases, according to the literature (Julian et al., 2007; WSU EB1240), literature (Julian et al., 2007; WSU EB1240), were: -were: -4.04.0°C for °C for calyx calyx red – first bloom phenophase intervalred – first bloom phenophase interval,, -2. -2.5°C for 5°C for bloomingblooming and and -2.0-2.0°C °C for for in the shuck – green fruit intervalin the shuck – green fruit interval. .

In this study, probability was defined as the ratio between the number In this study, probability was defined as the ratio between the number of years with unwanted events (floral organs damaged due to late of years with unwanted events (floral organs damaged due to late frost) and the total number of years under surveillance. frost) and the total number of years under surveillance.

This This climatic accidentsclimatic accidents occur when both the phenophase and the occur when both the phenophase and the critical temperature for this phenophase appeared simultaneouslycritical temperature for this phenophase appeared simultaneously..

These These non-mutually-exclusive events non-mutually-exclusive events are described are described by the following by the following formula (Hunt, 1986): formula (Hunt, 1986):

PP (SE) = (SE) = PP (F) * (F) * PP (T) (T) (1)(1) where:where:

- - PP(SE) = the probability of the simultaneous occurrence of the (SE) = the probability of the simultaneous occurrence of the two events, two events,

- - PP(F) = the probability of the phenophase occurrence, and (F) = the probability of the phenophase occurrence, and - - PP(T) = the probability of the occurrence of temperatures equal (T) = the probability of the occurrence of temperatures equal

to or lower than the critical temperatures during the same time to or lower than the critical temperatures during the same time periodperiod

• ThThe abovee above equation equation isis used used whenwhen the the simultaneous events outcomes simultaneous events outcomes (phenological dynamics and critical (phenological dynamics and critical temperature)temperature) are independentare independent;;

• TThere was nohere was no correlation between correlation between the onset of the onset of the the three phenophases three phenophases and the probability of occurrence of and the probability of occurrence of the critical minimum temperatures the critical minimum temperatures for the next 10-day periodfor the next 10-day period, after the , after the phenophase onsetphenophase onset

Because the events occur over a definite period of time, a 5-Because the events occur over a definite period of time, a 5-

day time interval was selected using the available data; that day time interval was selected using the available data; that

means 24 years of phenological observations and about 120 means 24 years of phenological observations and about 120

daily minimum air temperature values for each 5-day period, daily minimum air temperature values for each 5-day period,

measured over 24 years.measured over 24 years.

The daily minimum temperature data of each five-day The daily minimum temperature data of each five-day

interval between February 1interval between February 1stst, 1985, and May 31, 1985, and May 31stst, 2008, , 2008,

were condensed into probability functions for which the were condensed into probability functions for which the

assumption of normal distribution was validated by the assumption of normal distribution was validated by the

Shapiro-Wilk statistical test (SPSS 14.0, USA). Shapiro-Wilk statistical test (SPSS 14.0, USA).

These functions were used to calculate the probability of These functions were used to calculate the probability of

occurrence for temperatures lower than the phenophases occurrence for temperatures lower than the phenophases

critical values.critical values.

February through April period is considered of major importance for the onset of the fruit February through April period is considered of major importance for the onset of the fruit trees growing seasontrees growing season (Chmielewski et al., 2004): (Chmielewski et al., 2004):

Correlation between the length of the period from January 1Correlation between the length of the period from January 1stst to the date of onset of the to the date of onset of the each phenophase (‘Umberto’ cv.)each phenophase (‘Umberto’ cv.) and and the February through April mean air temperaturethe February through April mean air temperature

According to the collected data at Constanta weather station, an trend of increase in mean, minimum and maximum air temperature for the February through April months during the 1985 - 2008 period, was noticed only on polynomial trendline regression.

The was no statistically significant increase trend on linear regression.

TEMPERATURE DYNAMICSTEMPERATURE DYNAMICS

b) There was a b) There was a trend of increase trend of increase in in standard standard deviationdeviation of daily of daily maximum values maximum values of air of air temperaturetemperature in in FebruaryFebruary (0.55°C (0.55°C per decade)per decade)

a)a)

b)b)

FEBRUARYFEBRUARY

a) There wasa) There was no an increase trend no an increase trend for all the maximum, minimum and for all the maximum, minimum and mean temperatures during February mean temperatures during February (1985 – 2008) (1985 – 2008)

MARCHMARCH

a)a)

b)b)

a) There wasa) There was an an increase trendincrease trend for for all the maximum, all the maximum, minimum and meanminimum and mean temperatures temperatures during Marchduring March ((1.55°C per decade 1.55°C per decade for mean, 1.94°C for for mean, 1.94°C for maximum and only maximum and only 1.14°C for minimum 1.14°C for minimum temperaturetemperature

b) There was not a trend of b) There was not a trend of increase in standard deviation increase in standard deviation of daily values of air of daily values of air temperaturetemperature during Marchduring March

APRILAPRIL

b)b)

a)a)a) There wasa) There was no an increase trend no an increase trend for all the maximum, minimum and for all the maximum, minimum and mean temperatures during Aprilmean temperatures during April

b) There was a b) There was a trend of increase in trend of increase in standard deviationstandard deviation of daily minimum of daily minimum values of air values of air temperaturetemperature in Aprilin April ((0.41°C per decade0.41°C per decade))

Trend of spring average temperature in Romania (°C) during 1961-2007. Confidence level minimum 95% for hatch areas

Histograms of the onset of Histograms of the onset of calyx redcalyx red (a) (a),, first first bloom (b) bloom (b) and and end of petal fallend of petal fall ( (cc) )

phenophases phenophases in the ‘in the ‘UmbertoUmberto’ ’ apricotapricot cultivar cultivar at at ConstantaConstanta (19 (1985 - 200885 - 2008).).

a)a)b)b)

c)c)

The Shapiro-Wilk test value, T, - Calyx red T = 0.964 (P=0.627);- First bloom T = 0.960 (P=0.549); - In the shuck T = 0.970 (P=0.751).

The hypothesis of normal data distributioncould not be rejected

‘‘EMPIRICAL’ FREQUENCIESEMPIRICAL’ FREQUENCIES

In order to check the hypothesis that the phenological data can be approximated by the normal distribution

PHENOLOGICAL DYNAMICSPHENOLOGICAL DYNAMICS

Effect of the air Effect of the air temperature increasing temperature increasing trend was the earlier trend was the earlier onset of apricot tree onset of apricot tree vegetationvegetation

•‘‘Umberto’ apricot tree cultivar used to Umberto’ apricot tree cultivar used to

have its onset have its onset of calyx redof calyx red on April 12on April 12,, 25 25

years agoyears ago..• NNow it has its onset in vegetation on 26ow it has its onset in vegetation on 26thth

of March (17 days earlierof March (17 days earlier))

aa))

b)b)

Processing phenologProcessing phenologicalical data data

a) Time distribution of the a) Time distribution of the phenophase stage frequency phenophase stage frequency computed from onset dates computed from onset dates of each successive of each successive phenophasesphenophases:: calyx red, first calyx red, first bloom and petal fall (1985-bloom and petal fall (1985-2008) 2008)

b) Time distribution of the b) Time distribution of the phenophase stage probability phenophase stage probability computed from computed from mean date mean date and standard deviation of and standard deviation of each phenophase shown in each phenophase shown in frequency time distributionfrequency time distribution

MODEL OUTPUTMODEL OUTPUT

From cFrom collected ollected data data we we computed: computed:

a) probability of a) probability of phenophase phenophase occurence, occurence, PP(F);(F);

b) Probability of b) Probability of critical critical temperature temperature occurence, occurence, PP(T);(T);

Probability of two Probability of two simultaneous simultaneous events: events: PP(SE)=(SE)=PP(F)*(F)*PP(T). (T).

a)a)

b)b)

a)a) Time distribution of late Time distribution of late frost damages in terms of frost damages in terms of phenophase stage phenophase stage frequency. frequency. ‘Umberto’ apricot cv. at ‘Umberto’ apricot cv. at ConstantaConstanta (1985-2008) (1985-2008)

b) Time distribution of late b) Time distribution of late frost damages in terms of frost damages in terms of phenophase stage phenophase stage probability. probability. ‘Umberto’ apricot cv. at ‘Umberto’ apricot cv. at ConstantaConstanta (1985-2008) (1985-2008)

a)a)

b)b)

To check if the trend of apricot earlier To check if the trend of apricot earlier onset of vegetation may influence onset of vegetation may influence the late frost damage probability, the late frost damage probability, we divided the 24 years interval we divided the 24 years interval into three periods of 8 years (1985-into three periods of 8 years (1985-1992, 1993-2000, 2001-2008).1992, 1993-2000, 2001-2008).

a)a) Simulation - 1985-1992;Simulation - 1985-1992;b)b) Simulation - 2001-2008. Simulation - 2001-2008.

The maximum sum of The maximum sum of flower damages flower damages probability increased probability increased 7.6 times (11.4%).7.6 times (11.4%).

Simulation model to predict Simulation model to predict apricot phenological stagesapricot phenological stages In order to apply this study In order to apply this study conclusions, conclusions, to to the the other regions other regions from Romania, from Romania, with no with no

phenological observationsphenological observations (but with available temperature data), w (but with available temperature data), we developed a e developed a model to predictmodel to predict ‘Umberto’ cv. phenological stages using the correlation between‘Umberto’ cv. phenological stages using the correlation between thethe hourly temperature sums and the phenological observations that were available hourly temperature sums and the phenological observations that were available for Constanta apricot orchards (1985-2008)for Constanta apricot orchards (1985-2008)..

SumsSums of four temperature intervals of four temperature intervals, with different biological effect, with different biological effect (≥ 2°C and < 8°C, (≥ 2°C and < 8°C, ≥ 8°C and < 14°C, ≥ 14°C and < 20°C and ≥ 20°C)≥ 8°C and < 14°C, ≥ 14°C and < 20°C and ≥ 20°C), have been cummulated, have been cummulated from from February 1February 1stst to the date of phenophase estimation to the date of phenophase estimation. There were taken into . There were taken into consideration five day intervals.consideration five day intervals.

After the first five days, we made the first simulation with the equation (2) to After the first five days, we made the first simulation with the equation (2) to compute the date of each phenophase onset: compute the date of each phenophase onset:

y = a + b1*x1 + b2*x2 + b3*x3 + b4*x4y = a + b1*x1 + b2*x2 + b3*x3 + b4*x4 (2) (2) wherewhere:: y = the number of days from January 1 y = the number of days from January 1stst to the onset of the phenophase, to the onset of the phenophase,

x1x1, 2, 3, 4, 2, 3, 4 = the sum of hours with temperatures = the sum of hours with temperatures inside inside ≥ 2°C and < 8°C≥ 2°C and < 8°C and the other above intervals; and the other above intervals;

We used the equation 2 for the onset of each of the three phenophases under We used the equation 2 for the onset of each of the three phenophases under surveillance;surveillance;

We repet the computation for each other added five day intervals, and stop the We repet the computation for each other added five day intervals, and stop the simulation when the simulated date (y-days counted from January 1simulation when the simulated date (y-days counted from January 1stst) is very close ) is very close to the date resulted from cumulating the five day intervals (starting with the 1to the date resulted from cumulating the five day intervals (starting with the 1stst February);February);

We considered February 1We considered February 1stst as the starting date for calculating the sum of hours, as the starting date for calculating the sum of hours, because in January, in Romania, in all cases apricot trees are in the endodormancy because in January, in Romania, in all cases apricot trees are in the endodormancy period. period.

Model validationModel validation with with Constanta orchard dataConstanta orchard data

The correlation between the The correlation between the rerecordedcorded dates of onset of dates of onset of phenophases (y) and the phenophases (y) and the simulated onessimulated ones (x) is very (x) is very highly significanthighly significant

Simulation - 1985-1992;Simulation - 1985-1992;Simulation - 2001-2008.Simulation - 2001-2008.

CraiovaCraiova

The maximum sum The maximum sum of flower damages of flower damages probability increased probability increased 7.3 times (4.9%).7.3 times (4.9%).

ConclusionsConclusions•By thisBy this methodmethod one can be one can be computed the dynamics of computed the dynamics of probability of late frost damage occurring in apricot probability of late frost damage occurring in apricot orchards (‘Umberto’ and other 4 cultivars) at Constanta, orchards (‘Umberto’ and other 4 cultivars) at Constanta, South-Eastern Romania in the latest 25 yearsSouth-Eastern Romania in the latest 25 years;;

•In the 1985 – 1992 period the maximum sum of flower In the 1985 – 1992 period the maximum sum of flower damage probability was of only 1.5% (11-20 March) for damage probability was of only 1.5% (11-20 March) for ‘Umberto’‘Umberto’. In this case, the. In this case, the area area may may be characterized as be characterized as very favourable for growing apricotvery favourable for growing apricot;;

•16 years later (2001 – 2008) the maximum sum of flower 16 years later (2001 – 2008) the maximum sum of flower damage probability increased 7.6 times damage probability increased 7.6 times ((11.4%11.4%)). In the 1-. In the 1-5 April interval once in 8.8 years apricot flowers 5 April interval once in 8.8 years apricot flowers werewere damaged by late frostdamaged by late frost. In this case,. In this case, area area may bemay be described described with highwith high risk from this point of view. risk from this point of view.

AcknowledgmentThe work was carried out with the financial support of CNCIS,

Program Idei, project 1035/2007.