Transcript Document

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
Due to the changes of continental-temperate climate of
Romania, the late frosts affecting apricot trees by
damaging flowers are happening more often
Objectives

The purpose of this paper was to develop a
simulation model that estimates not only the
phenological dynamics until the end of blooming,
but also to estimate the occurrence probability of
climatic accidents caused by the late frosts in
apricot orchards from the South-Eastern part of
Romania during 1985 – 2008 period, under the
climatic change conditions
MATERIAL AND METHODS

A database of phenological observations concerning the
calyx red, first bloom and the end of petal fall (in the shuck),
in apricot (‘Prunus armeniaca L.’), ‘Umberto’, ‘C.R. 2-63’,
‘Sirena’, ‘Neptun’ and ‘Selena’ cultivars orchards, for a 24
year period (1985-2008), was used. The orchards
investigated were located within plain region of Constanta
(Valu lui Traian), South-Eastern Romania, with trees of
various ages.

Weather data: Mean, maximum and minimum daily
temperatures from the February 1st to the May 31st, during
24-year period interval were taken into consideration.




The critical temperatures for the three phenophases, according to the
literature (Julian et al., 2007; WSU EB1240), were: -4.0°C for calyx
red – first bloom phenophase interval, -2.5°C for blooming and -2.0°C
for in the shuck – green fruit interval.
In this study, probability was defined as the ratio between the number
of years with unwanted events (floral organs damaged due to late
frost) and the total number of years under surveillance.
This climatic accidents occur when both the phenophase and the
critical temperature for this phenophase appeared simultaneously.
These non-mutually-exclusive events are described by the following
formula (Hunt, 1986):
 P (SE) = P (F) * P (T)
(1)
where:
- P(SE) = the probability of the simultaneous occurrence of the
two events,
- P(F) = the probability of the phenophase occurrence, and
- P(T) = the probability of the occurrence of temperatures equal
to or lower than the critical temperatures during the same time
period
• The above equation is used when the
simultaneous events outcomes
(phenological dynamics and critical
temperature) are independent;
• There was no correlation between
the onset of the three phenophases
and the probability of occurrence of
the critical minimum temperatures
for the next 10-day period, after the
phenophase onset

Because the events occur over a definite period of time, a
5-day time interval was selected using the available data;
that means 24 years of phenological observations and
about 120 daily minimum air temperature values for each
5-day period, measured over 24 years.

The daily minimum temperature data of each five-day
interval between February 1st, 1985, and May 31st, 2008,
were condensed into probability functions for which the
assumption of normal distribution was validated by the
Shapiro-Wilk statistical test (SPSS 14.0, USA).

These functions were used to calculate the probability of
occurrence for temperatures lower than the phenophases
critical values.
February through April period is considered of major importance for the onset of the fruit
trees growing season (Chmielewski et al., 2004):
Correlation between the length of the period from January 1st to the date of onset of the
each phenophase (‘Umberto’ cv.) and the February through April mean air temperature
TEMPERATURE DYNAMICS
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.
FEBRUARY
a)
b) There was a
trend of increase
in standard
deviation of daily
maximum values
of air
temperature in
February (0.55°C
per decade)
a) There was no an increase trend
for all the maximum, minimum and
mean temperatures during February
(1985 – 2008)
b)
MARCH
a)
b) There was not a trend of
increase in standard deviation
of daily values of air
temperature during March
b)
a) There was an
increase trend for
all the maximum,
minimum and mean
temperatures
during March
(1.55°C per decade
for mean, 1.94°C for
maximum and only
1.14°C for minimum
temperature
APRIL
a)
a) There was no an increase trend
for all the maximum, minimum and
mean temperatures during April
b) There was a
trend of increase in
standard deviation
of daily minimum
values of air
temperature in April
(0.41°C per decade)
b)
Trend of spring average temperature in Romania (°C) during 1961-2007. Confidence level
minimum 95% for hatch areas
a)
‘EMPIRICAL’ FREQUENCIES
b)
c)
In order to check the hypothesis that the
phenological data can be approximated by the
normal distribution
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 distribution
could not be rejected
Histograms of the onset of calyx red (a), first
bloom (b) and end of petal fall (c)
phenophases in the ‘Umberto’ apricot cultivar
at Constanta (1985 - 2008).
PHENOLOGICAL DYNAMICS
Effect of the air
temperature increasing
trend was the earlier
onset of apricot tree
vegetation
•‘Umberto’ apricot tree cultivar used to
have its onset of calyx red on April 12, 25
years ago.
• Now it has its onset in vegetation on 26th
of March (17 days earlier)
a)
Processing phenological data
a) Time distribution of the
phenophase stage frequency
computed from onset dates
of each successive
phenophases: calyx red, first
bloom and petal fall (19852008)
b)
b) Time distribution of the
phenophase stage probability
computed from mean date
and standard deviation of
each phenophase shown in
frequency time distribution
MODEL OUTPUT
From collected
data we
computed:
a) probability of
phenophase
occurence, P(F);
b) Probability of
critical
temperature
occurence, P(T);
Probability of two
simultaneous
events:
P(SE)=P(F)*P(T).
a)
a) Time distribution of late
frost damages in terms of
phenophase stage
frequency.
‘Umberto’ apricot cv. at
Constanta (1985-2008)
b)
b) Time distribution of late
frost damages in terms of
phenophase stage
probability.
‘Umberto’ apricot cv. at
Constanta (1985-2008)
a)
To check if the trend of apricot earlier
onset of vegetation may influence
the late frost damage probability,
we divided the 24 years interval
into three periods of 8 years (19851992, 1993-2000, 2001-2008).
a) Simulation - 1985-1992;
b) Simulation - 2001-2008.
b)
The maximum sum of
flower damages
probability increased
7.6 times (11.4%).
Simulation model to predict apricot phenological stages
In order to apply this study conclusions, to the other regions from Romania, with no
phenological observations (but with available temperature data), we developed a
model to predict ‘Umberto’ cv. phenological stages using the correlation between
the hourly temperature sums and the phenological observations that were available
for Constanta apricot orchards (1985-2008).
 Sums of four temperature intervals, with different biological effect (≥ 2°C and < 8°C,
≥ 8°C and < 14°C, ≥ 14°C and < 20°C and ≥ 20°C), have been cummulated from
February 1st to the date of phenophase estimation. There were taken into
consideration five day intervals.
 After the first five days, we made the first simulation with the equation (2) to
compute the date of each phenophase onset:
y = a + b1*x1 + b2*x2 + b3*x3 + b4*x4
(2)
where: y = the number of days from January 1st to the onset of the phenophase,
x1, 2, 3, 4 = the sum of hours with temperatures inside ≥ 2°C and < 8°C and
the other above intervals;
 We used the equation 2 for the onset of each of the three phenophases under
surveillance;
 We repet the computation for each other added five day intervals, and stop the
simulation when the simulated date (y-days counted from January 1st) is very close
to the date resulted from cumulating the five day intervals (starting with the 1st
February);
We considered February 1st as the starting date for calculating the sum of hours,
because in January, in Romania, in all cases apricot trees are in the endodormancy
period.

Model validation with
Constanta orchard data
The correlation between the
recorded dates of onset of
phenophases (y) and the
simulated ones (x) is very
highly significant
Craiova
Simulation - 1985-1992;
Simulation - 2001-2008.
The maximum sum
of flower damages
probability increased
7.3 times (4.9%).
Conclusions
•By this method one can be computed the dynamics of
probability of late frost damage occurring in apricot
orchards (‘Umberto’ and other 4 cultivars) at Constanta,
South-Eastern Romania in the latest 25 years;
•In the 1985 – 1992 period the maximum sum of flower
damage probability was of only 1.5% (11-20 March) for
‘Umberto’. In this case, the area may be characterized as
very favourable for growing apricot;
•16 years later (2001 – 2008) the maximum sum of flower
damage probability increased 7.6 times (11.4%). In the 1-5
April interval once in 8.8 years apricot flowers were
damaged by late frost. In this case, area may be
described with high risk from this point of view.
Acknowledgment
The work was carried out with the financial support of CNCIS,
Program Idei, project 1035/2007.