Scientia Horticulturae 103 (2005) 361–379

www.elsevier.com/locate/scihorti

Temperature effects on flower formation

in saffron (Crocus sativus L.)
R.V. Molina, M. Valero1, Y. Navarro1,
J.L. Guardiola, A. Garcı́a-Luis*
Departamento de Biologı́a Vegetal, Universidad Politécnica de Valencia,
46071 Valencia, Spain
Accepted 9 June 2004

Abstract

The temperature conditions for shoot growth and flower formation were characterised for saffron
(Crocus sativus L.). Leaf withering occurred during late winter or spring depending on location, and
coincided with a rise in temperature. No growth was detectable in the buds during the first 30 days after
leaf withering, neither in underground corms nor in lifted corms incubated in the laboratory under
controlled conditions. Flower initiation occurred during the first growth stages of the buds. The optimal
temperature for flower formation was in the range from 23 to 27 8C, 23 8C temperature being
marginally better. To ensure the formation of a maximum number of flowers, the incubation at these
temperatures should exceed 50 days, although incubation longer than 150 days resulted in flower
abortion. Flower emergence required the transfer of the corms from the conditions of flower formation
to a markedly lower temperature (17 8C). Incubation of the corms after lifting at a higher temperature
(30 8C), reduced flower initiation and caused the abortion of some of the initiated flowers. No flowers
formed in corms incubated at 9 8C. A variable proportion (20–100%) of the corms forced directly at
17 8C without a previous incubation at 23–27 8C formed a single flower. The wide differences in the
timing of the phenological stages in different locations we found in this study seemed related to the
ambient temperature. Leaf withering was followed shortly by flower initiation, which occurred during
late spring or early summer as the rising temperature reached 20 8C. A long hot summer delayed flower
emergence which occurred in late autumn as the temperature fell to the range of 15–17 8C.
# 2004 Elsevier B.V. All rights reserved.

Keywords: Crocus sativus; Dormancy; Flower forcing; Flowering; Saffron; Temperature

* Corresponding author. Tel.: +34 96 3877411; fax: +34 96 3877419.

E–mail address: [email protected] (A. Garcı́a-Luis).
1
Formerly at Compañı́a General del Azafrán de España, Liria, Valencia, Spain.

0304-4238/$ – see front matter # 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.scienta.2004.06.005
362 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

1. Introduction

Saffron (Crocus sativus L.) is an autumn-flowering geophyte extensively grown in the

Near East and the Mediterranean basin since the Late Bronze Age (Zohary and Hopf, 1994;
Negbi, 1999). Its long scarlet stigmas were highly valued for flavouring foods and for
colouring them golden-yellow. They were also used for dying textiles. The appreciation for
saffron as a food additive continues today and it has become the world’s highest priced
spice (Winterhalter and Straubinger, 2000). There is also a long tradition of saffron use in
the traditional medicine of many cultures (Abdullaev, 1993; Ma et al., 2001). More
recently, there has been increasing interest in the biological effects of the components of
saffron and their potential medical applications, particularly those based on their cytotoxic,
anticarcinogenic and antitumour properties (Abdullaev and Frenkel, 1999; Fernández
Pérez and Escribano Martı́nez, 2000).
Despite these apparently bright prospects, there was a marked reduction in saffron
production during the last two decades in some of the traditional producing countries, as
Spain, Italy and Greece (Negbi, 1999). This reduction in production has been most marked
in Spain, a country rightly considered in the near past as the leading producing country in
the world, where the land area devoted to saffron cultivation fell from 2800 ha in 1990 to
under 600 ha in 1999 (Ministerio de Agricultura, Pesca y Alimentación, 2000). These
figures are well below the 47,000 ha devoted to saffron cultivation in Spain in the early
1970s and were accompanied by a parallel reduction in production from over 44 metric
tons in 1976 to under 6 metric tons in 1999. The main reason for this reduction in
production is that the technology of saffron production has not changed from the ancient
times. Increasing labour costs have turned saffron production unprofitable despite its high
market price. The picking of crocus flowers (0.25–2.5 million flowers ha 1, depending on
the productivity of the plantation) and the separation of the stigmata require an intensive
hand labour. Saffron flowering lasts only 2–3 weeks, and picking of the flowers is required
daily. A technological breakthrough to increase the profitability of saffron cultivation
would be the cultivation under controlled conditions. In this way, flowering period could be
extended, thus reducing the need of intensive labour. Further, it would be much easier to
mechanise blossom collection in containerised plants than in those grown in the soil.
Despite its importance, the flowering process in saffron has not been characterised
precisely. The life cycle of saffron is similar in all producing countries, but there are wide
differences in the timing of events (Alarcón Molina and Sánchez Requena, 1968; Wilkins,
1985; Botella et al., 2002). Flowering occurs during autumn (October–November), and is
followed by a vegetative stage throughout winter and the formation of replacement corms
at the base of the shoots. At the beginning of the dry season (April–May), the leaves
senesce and wither, and the bulbs go into dormancy. The transition from the vegetative to
the reproductive stage occurs shortly afterwards in the apex of the buds of underground
corms. This transition has been reported to start during March in Azerbaijan (Milyaeva and
Azizbekova, 1978; Azizbekova and Milyaeva, 1999), from March to April in Israel
(Greenberg-Kaslasi, 1991) and during July in Kashmir (Koul and Farooq, 1984).
Differences in corm size or seasonal variations have been considered the cause of
these differences in transition dates (Negbi, 1999). Flower formation is directly related
to corm size (Negbi et al., 1989; De Mastro and Ruta, 1993) and a quantitative relationship
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 363

between these two parameters was found (Negbi et al., 1989). However,
the role of ambient temperatures on flower bud differentiation and subsequent flowering
is largely unknown. Plessner et al. (1989) reported the formation of a similar number of
flowers in corms forced either under uncontrolled conditions (at around 15 8C) or in a
phytotron at a 17/12 8C (day/night) cycle. However, neither the temperature nor the
duration of corm storage before corm forcing was stated, apart from the rather imprecise
statement ‘‘under ambient room conditions (not controlled)’’. Benschop (1993) quoted
Le Nard to state that C. sativus corms can be stored at 30 8C and 80% relative humidity
for up to 8 months, thus retarding flowering. However, no quantitative data were
presented. Muñoz Gómez et al. (2002) stated that the storage of the corms at 30 8C for
45 days increased the number of flowers as compared to corms forced to sprout directly
at 17/10 8C after leaf withering. However, the number of flowers formed was very low
(under 0.3 flowers per corm).
In this study, we characterised the life cycle of saffron in several locations with different
climates. Also, we determined the influence of temperature on the transition from the
vegetative to the generative state, and on the dormancy of the corms.

2. Materials and methods

The corms used in this study were grown at Quero, Toledo, a traditional saffron
producing area of Spain. Shortly after leaf senescence (late May to early June) the plants
were lifted, the mother corm parts removed, and the new replacement corms separated and
graded. The clean replacement corms were then dipped in a prochloraz solution (0.1%) to
prevent Fusarium and Penicillium infestations and dried under forced ventilation for 5–7 h
to remove the surface water. The dry corms were kept under shelter during 3–5 days at
ambient temperature (17–20 8C) before the start of the experiments. Graded corms of the
desired size were placed on a mat of dry rock wool in plastic trays (0.35 m  0.5 m, and
9 cm height) and incubated in the dark at the desired temperature (1 8C). Temperature
and duration of incubation are given in Section 3 for every experiment. A humidity sensor-
controlled water atomizer kept the relative humidity during this storage at 85  2%.
Ventilation was provided twice a day to keep the CO2 concentration below 400 ppm. For
each treatment (a set of temperature and duration of incubation) three trays (80 bulbs each)
were prepared. Two trays were used for the visual inspection of time of flowering and
flower count and to harvest of saffron. The third tray was used for the measurement of
growth, the destructive characterisation of the flowering stage of the apex, to monitor
weight loss by the bulbs and to take samples for histology. Samples for these observations
consisted of ten corms selected at random.
After this incubation, the corms were covered with a 4–5 cm deep expanded clay (arlite)
layer, watered with 1/4 strength Hoagland solution and transferred to a dark room at 17 8C
for flower emergence. At the start of anthesis, light was provided by Pope FTD 58W/82
fluorescent lamps, with a photoperiod of 8/16 h (light/dark) and a photon flux density of
20 mmol m 2 s 1. In some experiments, prior to the forcing at 17 8C, the watered corms
were kept in the dark at an intermediate temperature (21 8C) for a variable length of time
(6–18 days).
364 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

Fig. 1. (A) Multiple shoot development from a single corm. The growing replacement corms formed at the base of
these shoots shall grow little and shall not flower the next season. (B) Replacement corm formed at the base of the
single flowering shoot present in the mother corm. This corm shall grow to reach 30–40 mm of diameter and shall
flower next season. Fibrous caulogenic roots (arrow) formed a root plate at the base of the mother corm. (C)
Mature replacement corm formed at the base of a non-flowering shoot with the tunic (the expanded basis of the
true and the sheathing leaves) removed to show the presence of a single apical dominant bud. (D) A mature
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 365

The stage of flower initiation and development was characterised using a numerical
scale (see Section 3) based on the stages of development described by Beyer (1942). For
histological observations, the growing point was removed with a minimal amount of corm
tissue, fixed and embedded in paraffin. Longitudinal median sections were cut at a
thickness of 10 mm, and stained with either the Schiff reagent or safranin and fast green
(Johansen, 1940). The significance of the environmental effects on flowering was tested by
means of an analysis of variance (ANOVA). Mean separation was done using a multiple
range test (Snedecor and Cochran, 1967).

3. Results

3.1. The structure of the corms

A replacement corm formed at the base of every shoot developed from the maternal
corm. The size and structure of these replacement corms depended on their number and
on flower formation. A high number of shoots per corm resulted in the formation of
several small-sized replacement corms which did not flower the following season
(Fig. 1 A). The number of shoots in the mother corms was usually much lower, ranging
from one (Fig. 1B) to three. The replacement corms formed at the base of non-
flowering shoots kept the apical meristem of the shoot, which became the dominant,
and usually the only one, sprouting bud the following season (Fig. 1C and E). The
apical meristem of the flowering shoots was lost (Fig. 1F), and the replacement corms
formed at the base of these shoots had two to three well developed axillary buds
(Fig. 1D). All these buds sprouted the following season and formed a new replacement
corm. This resulted in the increase in the number of corms. A sheath of cataphylls
protected all the main buds. These cataphylls grew ahead of the shoot apex, and pierced
the soil, thus protecting the shoot apex during emergence.
Replacement corms, either from flowering or non-flowering shoots, with a diameter
ranging from 25 to over 40 mm, were used in our experiments.

3.2. The influence of location on ontogenesis

The ontogenesis of C. sativus plants grown in the open in four locations of different
climates is shown in Fig. 2. The period without aboveground organs was much longer
in Jerez (circa 9 months; from March to the end of November) than in Albacete (4.5
months; from early June to mid-October) or Segovia (near 4 months). At Valencia, this

replacement corm formed at the base of a flowering shoot with the tunic removed to show three main big-sized
axillary buds, which shall sprout in the following season. The smaller axillary buds at more basal positions in the
corm (seen as small grey dots in the photograph) usually remain quiescent. (E) Longitudinal section of a corm
formed at the base of a non-flowering shoot. The apical meristem has formed a protective sheath with several
cataphylls. It shall become the apex of the replacement corm. Numerous vascular connections connect the mother
corm and the replacement corm (arrows). (F) Longitudinal section of a replacement corm formed at the base of a
flowering shoot. Flower formation resulted in the loss of the apical meristem of the shoot. Two axillary buds
protected by a sheath of several cataphylls (arrows) have done a significant growth.
366 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

Fig. 2. Ontogenesis of C. sativus plants in four locations in Spain. Phenological data are mean from observations
carried-out during three consecutive years on plantings started with a uniform batch of corms. Monthly maximum,
average and minimum air temperatures (means for the period 1961–1990) were obtained from the Instituto
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 367

Fig. 3. (A) Apical bud growth and organogenesis in the apex during the summer in corms grown at Albacete, lifted
on 23 June (23 days after leaf withering) and incubated under controlled conditions at 25 8C with 85% relative
humidity (open symbols), and in corms grown at Quero (Toledo) and sampled periodically during summer (closed
symbols). (B) Size of the shoot apex in the corms from Quero. For the measurements performed before August, the
standard errors were smaller than symbols size; in later dates, smaller than 4% of the value of the parameter.

period had an intermediate duration (around 7 months, from early April to early
November).
There seemed to be a close relationship between the air temperature and the timing of
the phenological events (Fig. 2). The warmer winter climates of Jerez and Valencia resulted

Nacional de Metereologı́a. The timing of flower differentiation was determined from the observation of the dominant
bud from corms 30–40 mm diameter. Symbols: B, beginning of the formation of the replacement corm; S, senescence
of the leaves; I, flower initiation in the dominant shoot apex; R, emergence of the caulogenic roots; F, anthesis.
368 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

Fig. 4. Low magnification photographs of the naked shoot apex (having the sheath of cataphylls removed) at
different developmental stages, and flower scores used in the present work. (A) Resting bud on 25 June. No lateral
organs were initiated at this time (flower score, 1). (B) Dome-shaped shoot apex (13 July) with leaf primordia
initiated at the flanks of the meristem (flower score, 2). (C) Shoot apex on 20 July. The base of the dome-shaped
meristem was covered by the developing leaf primordia (flower score, 3). (D) Shoot apex on 25 July, at the time the
bracts were initiated at the edge of the meristem (flower score, 4). (E) Stamen initiation (6 August; flower score, 6).
(F) Bud with three developing flowers (14 August). The stamens were much longer than the leaves, a consequence
of the hysteranthy of this species. At the dorsal side of the stamens, the first whorl of tepals was already initiated in
two of the flowers (arrow; flower score, 8). (G) Shoot apex with two developing flowers on 20 August. The
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 369

in an early withering of the leaves and the shortening of the vegetative growth period. The
earlier rise in temperature during early spring at these locations, accelerated flower
initiation, but the high autumn temperatures delayed root formation and flowering. In
all locations, flower differentiation occurred during late spring or early summer (early June
to late July), when the mean air temperature rose to about 20 8C. Anthesis occurred in
autumn, when the mean air temperature fell to 15–17 8C (Fig. 2). Flowering was always
preceded by root formation.

3.3. Shoot growth and flower initiation

The time-course of shoot growth and flower differentiation in the corms during summer
is presented in Fig. 3. The graphs show combined data from corms lifted after leaf
senescence and incubated at 25 8C, and from corms kept in situ on the field at Quero, as no
significant differences were observed in shoot development and the timing of
morphogenesis. The numerical scale of flower differentiation we used was based on
the stages of meristem development as defined by Beyer (1942). The main stages are
illustrated in Figs. 4–6.
No change in the size of the bud (the length of the outermost cataphylls; Fig. 3A) or that
of the naked shoot apex (without cataphylls; Fig. 3B) was observed from the time of corm
lifting in early June to early July, some 40 days after leaf senescence. During this time, the
buds made no growth and seemed to be dormant. No flower initials were present in the
resting buds (flower score, 1; Figs. 4A, 5A and 6A). By early July, the apex become dome-
shaped (flower score, 2; Figs. 4B and 6B). The increase in size of the apex was shortly
followed by the formation of leaf primordia (flower score, 3; Figs. 4C, 5B and 6C). Shortly
afterwards (16 July) the sheathing leaves started to grow at a faster rate than the shoot apex
(Fig. 3A), piercing through the soil and protecting the growth of the young shoot and the
scape. In succession followed the bract primordia stage (flower score, 4; Figs. 4D and 5C),
the formation of stamen primordia (flower score, 6; Figs. 4E and 5D), the initiation of the
perianth (flower score, 8; Figs. 4F, 5E and 6D) and the formation of gynoecium (flower
score, 10 and higher; Figs. 4G and 6E). All the flower parts were already differentiated by
the end of August (Fig. 4G and H). At this time, the sheathing cataphylls where 7–9 mm
long (Fig. 3A).
The repeated observation of this sequence of flower morphogenesis in the same location
during two consecutive years showed a difference in 12 days to attain the formation of the
gynoecium stage.

3.4. Temperature requirements for flower formation

The influence of a constant temperature regime on flower formation was determined

using corms lifted immediately after leaf withering (early June). These corms were

gynoecium was already initiated (arrow; flower score, 10). (H) Further growth of the flower parts, ahead of the
leaves. (I) The gynoecium had reached one half of the length of the stamens. The developing style and stigmata
showed a typical reddish colour (darker in the photograph). Symbols: b, bract; g, gynoecium; l, leaf; st, stamen; t,
tepal. Scale bars, 0.5 mm.
370 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

Fig. 5. Scanning electron microscope view of the naked shoot apex (with the sheath of cataphylls removed)
showing the main stages of organ formation. (A) Resting bud (flower score, 1). (B) Leaf differentiation. The base
of the leaf primordia almost encircled the shoot apex (flower score, 3). (C) Bract formation. The basal part of the
shoot was protected by the tightly packed leaf primordia (flower score, 4). (D) Bud with the stamens being initiated
(flower score, 6). (E) Bud with two developing flowers. In the oldest flower the six tepals were already
differentiated (only five of them are shown in the picture as the sixth was removed during the preparation of
the sample). In the younger flower, the first whorl of tepals was being initiated at the dorsal side of the stamen
(arrows; flower score, 8). (F) Bud with two developing flowers. The bilobed stamens are easily distinguished from
the arrowhead pointed leaves. At this stage, the gynoecium was being initiated (not seen in the picture; flower
score, higher than 8). Symbols: s, cataphyll scar; other symbols as in Fig. 4.
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 371

Fig. 6. Longitudinal sections of the shoot apex during the main stages of organ formation. (A) Sampled on 7 June.
No lateral organs were differentiated. The apex was protected by several cataphylls (flower score, 1). (B) Sampled
on 5 July. The shoot apex was dome-shaped, but no lateral organs were initiated (flower score, 2). (C) Sampled on
12 July. The shoot apex had increased in size and the leaves differentiated at the flank of it (flower score, 3). (D)
Sampled on 11 August, with the tepals differentiated (flower score, 8). (E) Sampled on 20 August, showing the
initial stages of gynoecium formation (flower score, 10). Symbols: as in Figs. 4 and 5. Scale bar: 0.5 mm.
372 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

Fig. 7. The influence of incubation temperature on flower formation. The corms were incubated for 100–130 days
at the temperature indicated before being transferred to 17 8C for flower emergence. Number of flowers expressed
as a percentage of the flowers formed at 25 8C. Data from three separate experiments (indicated with different
symbols). The incubation at 25 8C was performed in the three experiments. The number of flowers formed per
corm at this temperature ranged from 2.4 to 2.8. The standard error was less than 0.12 flowers per corm (5% in the
units of the graph).

incubated at the desired temperature (in the range 9–30 8C) for 100–130 days, and then
forced to sprout at 17 8C. This temperature is close to the optimum for flower emergence as
determined in prior experiments (evidence not presented). The results from three separate
experiments carried-out in different years are presented in Fig. 7. The number of flowers
formed was maximal for corms incubated at 23–27 8C. With the experimental material we
used, the corms incubated at any temperature within this range for 90–150 days, formed
between two and three flowers depending on the experiment. The growth of the shoot apex
and floral morphogenesis were slightly faster, and anthesis occurred earlier, when the
incubation was performed at the lowest of these temperatures (23 8C; Table 1). The corms
held at this temperature also flowered slightly but significantly earlier. Within this range of
temperature, there was no significant effect of the temperature of incubation either in
flower number or in saffron production (Table 1).
Incubation at a temperature outside the range of 23–27 8C resulted in a reduction in
flower number per corm (Fig. 7). This reduction was more marked at a higher (30 8C) than
at a lower temperature (21 8C). At 21 8C, the number of flowers formed was close to that
recorded at the optimal temperatures. A further reduction in the temperature of incubation
resulted in a marked reduction in flower formation. No flowers formed in the corms
incubated at 9 8C. A variable proportion (20–100%, depending of the experiment) of the
corms incubated directly at 17 8C after lifting formed one single flower. This might indicate
a marginal (and variable) effect of this temperature on flower formation. All the batches of
corms used in our experiments were inspected under the dissecting microscope before
incubation and we found no evidence of flower initiation at the time the incubation was
started (the initial flowering score was invariably 1).
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 373

Table 1
The influence of incubation temperature on shoot growth and floweringa
Parameter Temperature of incubation
23 8C 25 8C 27 8C
Shoot apex length (mm)b 16.9 a 8.8 b 6.3 c
Flower stageb 8.9 a 8.0 b 7.0 c
Days to flowerc 123 a 128 b 141 c
Flowers per corm 2.2 2.2 2.3
Saffron per flower (mg) 10.3 10.2 10.3
a
Values in the same row with different letters differ significantly at P  0.05.
b
Measured after 77 days incubation.
c
For corms incubated for 102 days at the temperature indicated prior to forcing at 17 8C.

The effect of incubation duration on flowering was investigated for corms incubated
either at 25 or 30 8C (Fig. 8). Flower formation was higher (P  0.01) at 25 8C than at 30 8C
(Fig. 8). The maximum number of flowers was obtained al 25 8C with an incubation lasting
from 90 to 150 days (Fig. 8). A 180-day-long incubation resulted in flower abortion.
Incubation at 30 8C only resulted in flower formation when it lasted from 90 to 120 days
(Fig. 8). A longer incubation at 30 8C suppressed flower formation.
The visual inspection of the growing shoots after incubation and prior to flower forcing
at 17 8C demonstrated that the smaller number of flowers formed in the corms incubated at
30 8C resulted from the combination of two factors; the smaller number of flowers initiated
at 30 8C (as compared to corms incubated at 25 8C) and the abortion of some of these

Fig. 8. The influence of the duration of incubation at 25 8C (closed signs) and 30 8C (open signs) on flower
formation in corms lifted on 9 June (flowering score, 1) immediately after leaf withering. The controlled
experimental conditions were set-up on 15 June. After the time of incubation indicated, the corms were forced to
flower at 17 8C.
374 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

Table 2
Number of flowers formed per corm as determined from the visual inspection of the shoots at the end of the
incubation at high temperature (25 or 30 8C)
Temperatures of incubation and parameter Duration of incubation
59 91 115 150 178
Incubated at 25 8C
Flowers formed 2.2 2.7 2.6 2.3 2.4
Flowers aborted 0.1 ns 0.2 ns 0.3 ns 0.3 ns 2.2**
Incubated at 30 8C
Flowers formed – 2.0 2.0 1.7 1.8
Flowers aborted – 0.5* 0.4* 1.4** 1.4**
The number of flowers aborted is the difference to the number of flowers reaching anthesis after forcing the corms
at 17 8C.
ns: the difference to zero was not significant.
*
Significant at P  0.05.
**
Significant at P  0.01.

flowers (Table 2). No abortion of flowers occurred in corms incubated at 25 8C unless the
period of incubation extended beyond 150 days (Table 2). In these corms, the number of
flowers counted at the end of incubation did not differ significantly from the number of
flowers reaching anthesis (the difference between these two counts, which is the number of
flowers aborted, did not differ from zero).

3.5. The effect of a post-incubation at 21 8C

Vigorous root growth occurred when the corms were transferred to the low
temperature conditions (17 8C) for flower emergence. This root growth was caused by
the combination of the lowering in temperature and of the watering of the corms. Root
formation also occurred in non-watered corms after being covered with the arlite layer,
but root growth was much weaker. The aims of this experiment, in which a short
incubation at an intermediate temperature (21 8C) was performed before forcing the
corms at 17 8C, were to determine the influence of this intermediate temperature on
root growth and its effect on flower formation. The first of the goals could not be
accomplished. Root growth was apparently more vigorous at 21 8C than at 17 8C.
However, at both temperatures the vigorously growing roots penetrated the rock wool
and made the quantitative measurements impracticable.
The effect of this intermediate temperature (21 8C) incubation on flower formation
and characteristics in corms incubated for a different length of time at 25 8C is
presented in Table 3. As expected from the results presented above, the duration of the
incubation at 25 8C in the range of 77–104 days used in this experiment had only a
marginal effect on flower number. Even so, the number of flowers formed per corm was
slightly higher (P  0.01) with 104 days incubation (2.4 flowers per corm for the mean
of the three durations of incubation at 21 8C) than with 77 days (2.1 flowers per corm).
The incubation at 21 8C had no effect either in flower number or in the time of
flowering, but significantly increased flower size (saffron yield). This effect was already
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 375

Table 3
The influence of a post-incubation at 21 8C on flower formation and saffron yield in corms incubated at 25 8C for
77 and 104 days
Days of incubation Flowers Saffron per Days to flower
25 8C 21 8C per corm flower (mg) (from 22 June)

77 0 2.3 9.9 120

77 12 2.0 10.6 120
77 18 2.0 10.3 120
ns P  0.10 ns
104 0 2.5 10.2 a 129
104 6 2.2 11.6 b 129
104 16 2.5 11.5 b 130
ns P  0.05 ns
Corms lifted on 5 June (phenological stage of the apex, 1). Controlled conditions of incubation started on 22 June.

maximal after 6 days of incubation. It was more marked in the corms incubated at 25 8C
for a longer time (Table 3).

3.6. The influence of the time of corm lifting

The corms lifted before 20 July formed few flowers when forced directly at 17 8C. The
number of flowers formed with a direct forcing at 17 8C increased slightly but significantly
at later dates, as the corms were lifted at a more advanced stage of flower differentiation
(Fig. 9). The number of flowers per corm was close to one in the corms lifted on 15 August,

Fig. 9. The influence of incubation at 25 8C on flower formation in corms lifted at different stages of
morphogenesis. The corms were lifted at the dates indicated in the figure (flower score at the time of corm
lifting shown encircled), and forced to flower either directly at 17 8C (open triangles) or after being kept at 25 8C
until 10 September (filled circles).
376 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

around 46 days after leaf senescence. At this date, the apical bud of the corms had reached
the stage of stamen formation (flower score, 7; Fig. 9). The number of flowers formed
increased markedly when the lifted corms were incubated at 25 8C until 10 September. All
the corms incubated in this way were therefore kept at a high temperature for 103 days after
the withering of the leaves and before flower formation. This figure results from adding to
the time of incubation at 25 8C in the laboratory the time the corms were kept in the field
under the uncontrolled but hot June and July conditions. Flower formation was
significantly less in the corms lifted from 20 July to 6 August than in those lifted at earlier
and later dates. During this period of time (20 July to 6 August) the bracts and the stamens
were being initiated in the corms (Fig. 4).

4. Discussion

At the time of leaf withering, which occurred during late spring or early summer
after the winter vegetative growth period, the meristems of the replacement corms were
vegetative as ascertained by the morphological observations (Figs. 4–6). Flower
initiation occurred during summer, as reported to occur in several spring-flowering
bulbous plants, for example, Crocus vernus and Crocus flavus (Wilkins, 1985),
Hyacinthus, Iris and Tulipa (Hartsema, 1961; Le Nard and De Hertog, 1993). The
optimum temperature for flower initiation and development was in the range of 23–
27 8C (Fig. 7). At variance with the above-mentioned spring-flowering species, which
require a warm–cold–warm temperature sequence to flower (Wilkins, 1985), C. sativus
flowered best in a warm–intermediate temperature sequence. The transfer of the corms
after flower initiation at a temperature lower than 15–17 8C resulted in a reduction in
flower formation (Molina et al., in press).
At the optimal temperature for shoot growth and flower development (23–27 8C; Fig. 7),
apical bud growth started about 40 days after the withering of the leaves (Fig. 3). Flower
formation followed shortly afterwards, and the first stages of flower initiation were already
discernible by early August (Figs. 4–6). The release of bud dormancy, and hence the time
needed for flower formation, could be accelerated by curing the Crocus corms at 30 8C for
a short time (Molina et al., in press). A similar response was reported for the bulbs of Tulipa
(De Hertog et al., 1983), Freesia (Imanishi, 1993), Hyacinthus, Iris and Muscari (Le Nard
and De Hertog, 1993). While 20 days incubation at 30 8C was effective to release bud
dormancy, a longer incubation at this temperature (60 days) resulted in a reduction both in
flower initiation and in flower development (Figs. 7 and 8). Also, an incubation longer than
150 days at the optimal temperature range (23–27 8C) for flower initiation resulted in the
abortion of the flowers (Fig. 8 and Table 2). Optimal flower formation required incubation
for more than 60 days at the temperature range mentioned earlier (Fig. 8), followed by the
transfer of the initiated meristems to a lower temperature. After 60 days of incubation, the
meristems had formed the bract and in some of them the stamen could be discernible
(Figs. 4 and 5). This stage is somewhat earlier than the perianth formation stage
recommended for the low temperature forcing of the spring-flowering Crocus (Benschop,
1993). The optimal temperature for flower emergence was in the range 15–17 8C (evidence
not presented; Plessner et al., 1989).
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 377

While the experiments discussed earlier demonstrated that a sufficiently long

incubation at high temperature (23–27 8C) was needed to ensure the formation of a
maximum number of flowers, the flowering behaviour of corms lifted at different times
showed that additional factors were also affecting flower formation. The behaviour of
the corms lifted during the bract and the stamen differentiation stages, which formed
markedly less flowers (P  0.01) than corms lifted either during earlier or later stages
(Fig. 9) supports the saffron growers believe that disturbing the corms during flower
initiation interferes with flower formation. The most obvious explanation for this
behaviour, a putative stress resulting from the bright summer sunlight heating the lifted
corms, may not be the reason. We lifted the corms before sunrise, and kept them
protected under shelter during manipulation.
Incubation at an intermediate temperature prior to cool storage has been demonstrated
to increase flower quality in Tulipa and other forced bulbous plants (Hartsema, 1961). In C.
sativus, the incubation of the watered bulbs at 21 8C prior to forcing at 17 8C resulted in a
significant increase in flower size (Table 3). Since this incubation enhanced root growth
(quantitative data not available), we may speculate that root-produced cytokinins may have
increased flower growth. Azizbekova et al. (1978) reported an effect of exogenously
applied kinetin on shoot and flower growth.
Bulb and corm size is a major factor to determine the capacity of bulbous plants to
flower (Le Nard and De Hertog, 1993). A tight relationship has been demonstrated
between corm size and flowering in C. sativus (Negbi et al., 1989; De Mastro and Ruta,
1993). Negbi (1999) speculated that differences in corm size could be responsible for
the wide differences reported in the time of flower initiation and flowering in different
places (Milyaeva and Azizbekova, 1978; Koul and Farooq, 1984; Greenberg-Kaslasi,
1991). The wide differences we found at different locations both in the timing of flower
initiation and in the time of flowering in plantings arising from a uniform batch of
corms (Fig. 2), pointed-out to climatic factors as the main cause of these differences.
The flowers were initiated as the mean air temperature increased during late spring
above 20 8C (Fig. 2). The shoots remained underground during the hot summer, to
pierce the soil and flower when the temperature fell below 15 8C. Hence, the warmer
the climate during the winter months, the earlier was flower initiation and the later the
time of flowering. The time-course for flower initiation was similar for corms kept in
the orchard at Albacete and Quero and those incubated at the laboratory at the optimum
temperature as determined in this study (Fig. 3). This demonstrated that the natural
weather conditions during early summer in the main saffron producing area of Spain
(Castilla-La Mancha) were close to optimal for flower initiation. Further, the vegetative
growth of the plant lasted from mid-October to the end of May. This long period of
photosynthetic activity (over 7 months as compared to 3–4 months in warmer climates)
allowed the complete development of the replacement corms.

Acknowledgements

This research has been financed through a research contract with Compañı́a General del
Azafrán de España. We appreciate the technical support from the staff of this company.
378 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379

Especially, we thank to Mr. Agustin Escandón, general manager of the company, for
granting permission to publish these results. We thank Dr. A.M.M. Duarte for the
translation of the Russian papers.

References

Abdullaev, F.I., 1993. Biological effects of saffron. Biofactors 4, 83–86.

Abdullaev, F.I., Frenkel, G.D., 1999. Saffron in biological and medical research. In: Negbi, M. (Ed.), Saffron:
Crocus sativus L.. Harwood Academic Publishers, Australia, pp. 103–114.
Alarcón Molina, J., Sánchez Requena, A., 1968. El Azafrán: Hoja divulgadora 13-68-H. Ministerio de
Agricultura, Madrid.
Azizbekova, N.S.H., Milyaeva, E.L., 1999. Saffron cultivation in Azerbaijan. In: Negbi, M. (Ed.), Saffron: Crocus
sativus L.. Harwood Academic Publishers, Australia, pp. 63–71.
Azizbekova, N.S.H., Milyaeva, E.L., Lobova, N.V., Chailakyan, M.K.H., 1978. Effects of gibberellic acid and
kinetin on formation of flower organs in saffron crocus. Sov. Plant Physiol. 25, 227–233.
Benschop, M., 1993. Crocus. In: De Hertog, A., Le Nard, M. (Eds.), The Physiology of Flower Bulbs. Elsevier,
Amsterdam, (Chapter 19), pp. 257–283.
Beyer, J.J., 1942. De terminologie van de bloemaanleg der bloembolgewassen, vol. 46. Mededeelingen van de
Landbouwhoogeschool, Wageningen, pp. 1–17.
Botella, O., de Juan, A., Muñoz, M.R., Moya, A., López, H., 2002. Descripción morfológica y ciclo anual del
azafrán (Crocus sativus L.). Cuadernos de Fitopatologı́a 71, 18–28.
De Hertog, A.A., Aung, L.H., Benschop, M., 1983. The tulip: botany, usage, growth and development. Horticult.
Rev. 5, 45–125.
De Mastro, G., Ruta, C., 1993. Relation between corm size and saffron (Crocus sativus L.) flowering. Acta
Horticult. 344, 512–517.
Fernández Pérez, J.A., Escribano Martı́nez, J., 2000. Biotecnologı́a del azafrán. Ediciones de la Universidad de
Castilla-La Mancha, Cuenca.
Greenberg-Kaslasi, D., 1991. Vegetative and reproductive development in the saffron crocus (Crocus sativus L.).
M.Sc. thesis, The Hebrew University of Jerusalem (Quoted by Negbi, 1999).
Hartsema, A.H., 1961. Influence of temperatures on flower formation and flowering of bulbous and tuberous
plants. In: Ruhland, W. (Ed.), Handbuch der Pflanzenphysiologie, vol. 16. Springer Verlag, Berlin, pp. 123–
167.
Imanishi, H., 1993. Freesia. In: De Hertog, A., Le Nard, M. (Eds.), The Physiology of Flower Bulbs. Elsevier,
Amsterdam, (Chapter 21), pp. 285–296.
Johansen, D.A., 1940. Plant Microtechnique. McGraw-Hill, New York.
Koul, K.K., Farooq, S., 1984. Growth and differentiation in the shoot apical meristem of the saffron plant (Crocus
sativus L.). J. Indian Bot. Soc. 63, 153–160.
Le Nard, M., De Hertog, A., 1993. Bulb growth and development and flowering. In: De Hertog, A., Le Nard, M.
(Eds.), The Physiology of Flower Bulbs. Elsevier, Amsterdam, (Chapter 4), pp. 29–43.
Ma, X.Q., Zhu, D.Y., Li, S.P., Dong, T.T.X., Tsim, K.W.K., 2001. Authentic identification of stigma croci (stigma
of Crocus sativus) from its adulterants by molecular genetics analysis. Planta Med. 67, 183–186.
Milyaeva, E.L., Azizbekova, N.S.H., 1978. Cytophysiological changes in the course of development of stem
apices of saffron crocus. Sov. Plant Physiol. 25, 227–233.
Ministerio de Agricultura, Pesca y Alimentación, 2000. Anuario de Estadı́stica Agroalimentaria 1999. Madrid.
Molina, R.V., Valero, M., Navarro, Y., Garcı́a-Luis, A., Guardiola, J.L. The effect of time of corm lifting and the
duration of incubation at inductive temperature on flowering in the saffron plant (Crocus sativus L.). Scientia
Horticult., in press.
Muñoz Gómez, R.M., de Juan Valero, A., Botella Miralles, O., Moya Aparicio, A., 2002. Efectos del
mantenimiento a elevadas temperaturas y del etefón sobre la etapa vegetativa y la producción de flores y
cormos hijos en el azafrán (Crocus sativus L.). ITEA 98, 200–212.
Negbi, M., 1999. Saffron cultivation: past, present and future prospects. In: Negbi, M. (Ed.), Saffron: Crocus
sativus L.. Harwood Academic Publishers, Australia, pp. 1–18.
 R.V. Molina et al. / Scientia Horticulturae 103 (2005) 361–379 379

Negbi, M., Dagan, B., Dror, A., Basker, D., 1989. Growth, flowering, vegetative reproduction, and dormancy in the
saffron crocus (Crocus sativus L.). Isr. J. Bot. 38, 95–113.
Plessner, O., Negbi, M., Ziv, M., Basber, D., 1989. Effects of temperature on the flowering of the saffron crocus
(Crocus sativus L.): induction of hysteranthy. Isr. J. Bot. 38, 1–7.
Snedecor, G.W., Cochran, W.G., 1967. Statistical Methods. The Iowa University Press, Ames, pp. 65–99.
Wilkins, H.F., 1985. Crocus vernus: Crocus sativus. In: Halevy, A.H. (Ed.), Handbook of Flowering, vol. 2. CRC
Press, Boca Raton, FL, pp. 350–355.
Winterhalter, P., Straubinger, M., 2000. Saffron: Renewed interest in an ancient spice. Food Rev. Int. 16, 39–59.
Zohary, D., Hopf, M., 1994. Domestication of Plants in the Old World, second ed. Clarendon Press, Oxford.