Seasonal activity and reproductive biology of the ground beetle Carabus dufouri (Coleoptera: Carabidae)

This paper concerns the life-cycle of Carabus dufouri Dejean 1829, one of the most representative species of Carabus in the south of the Iberian Peninsula. The study is based on data of the annual activity patterns in the natural habitat, on anatomical ob­ servations related to the sex and age of the specimens, on the reproductive condition of females and, finally, on the results of labora­ tory rearing experiments carried out to study the oviposition patterns and the course of development of immature stages. The results indicate that C. dufouri shows the annual rhythm of autumn breeders. However, the rhythm may also be related to the winter-breeder type ofNorth Africa.


INTRODUCTION
Organisms in their habitats are confronted with sea sonal fluctuations of physical and biotic factors.Conse quently, in order to survive, their development, reproduc tion period, and population dynamics have to be adapted to these environmental fluctuations.Adaptation of species life cycle to the local environment involves a series of physiological and behavioural features, including growth rate, annual number of generations, dormancy, adult lon gevity and synchronization between the reproduction pe riod and environmental conditions, fecundity, optional polymorphism and population dynamics (Neumann, 1986).
Carabid beetles are a group of insects sufficiently wellknown as to allow the study of their life-cycles in relation to environmental factors.The copious literature published on this subject in the past few years (i.e., Kurka, 1972;Hůrka, 1973Hůrka, , 1986;;Thiele, 1977;Paarmann, 1979aPaarmann, , b, 1990Paarmann, , 1994;;Den Boer & Den Boer-Daanje, 1990), indi cates that the biology of carabid species respond on one hand to their habitat peculiarities, and on the other to ad aptations evolved in their evolutionary history.
It is generally accepted that in the Temperate zones the life cycles of ground beetles are governed by annual os cillations in climatic conditions.Also, phylogenetically closely related species show basically the same develop mental type, and in the different evolutionary trends a similar spectrum of reproductive rhythms can be found, suggesting a process of convergent evolution (Paarmann, 1979a).Thus, the conditions in each environment can re sult in the existence of a given developmental cycle, rather than affecting its evolution (Hůrka, 1986;Cárdenas etal., 1996).
Most studies of this kind have been done in the Tem perate zone of Central Europe and some also in North Af rica (Paarmann, 1975), but in transitional zones, such as the Iberian Peninsula, they are scant (de los Santos et al., 1985).Therefore, we considered that a research pro gramme designed to study the predominant annual life cy cles in meridional Europe was necessary to understand the evolution and relationships between the different re productive rhythms of carabids.In order to fill in this lack of information we started to work on the life-cycle of Ibe rian carabid species (Cárdenas & Bach, 1992a, b;Cárde nas, 1994;Cárdenas & Hidalgo, 1995;Cárdenas et al., 1996); we followed the criteria provided by Paarmann (1975), Thiele (1977) and Makarov (1994).
This paper investigates seasonal activity, age structure, duration and survival of immature stages of Carabus du fouri Dejean 1829.
Most Central European Carabus species are "autumnbreeders" (Thiele, 1977), while the North African Carabini studied are mostly "winter-breeders", showing a temperature-photoperiodic influence on the gonad dor mancy (Paarmann, 1979b).The cycles of "autumnbreeders" and "winter-breeders" represent, respectively, the northern and southern variants of a type of annual rhythm characteristic of the Temperate zone, which should find their optimal display in the Mediterranean ecosystems.Therefore, a series of studies have been started on reproductive biology of the Carabus species whose distribution area also includes the south of the Ibe rian Peninsula.

Species
According to Deuve (1994), Carabus dufouri is part of the Mesocarabus Thomson, 1875 subgenus which constituted a Tyrrhenian line widely distributed in Central and South Europe, reaching the north of Africa (Morocco), and having its greatest representation in the Iberian Peninsula with 3 species and 42 subspecies (sensu Forel & Leplat, 1998).In spite of this diversi fication, only C. dufouri has colonized the southern Iberian re gion to the south of the Guadalquivir River and it may be considered as a replacement species of C. lusitanicus of which different subspecies occupy the rest of the Peninsular area (Toulgoet & Lassalle, 1983).C. dufouri may be found at very different altitudes (from low altitude to 2,500 m) in open areas or in forests so it can be con sidered to be an eurytopic species.

The study area
The sampling was done in the Subbéticas Mountains Natural Park, which is integrated into the Bética Chain of mountains which shape the South of the Iberian Peninsula and are a group of massifs of middle altitude (1,000-1,200 m as middle term) of a calcarean nature that form a karstic landscape of great ecologi cal interest.They are surrounded by cultivated lowland (olive groves).The dominant vegetation is typical of a Mediterranean thick mixed forest, the Paeonio coriaceae-Quercetum rotundifoliae association, with a high level of degradation where bushes and pastures prevail.The climate is of the Mediterranean type, but with a clear Continental influence, belonging to a Mesomediterranean bioclimatic belt but, in the highest altitudes it may be considered as a Supramediterranean belt (Rivas-Martínez, 1987) (Fig. 1).

Annual activity pattern
The annual activity pattern of C. dufouri was recorded in a typical mediterranean mixed forest (La Nava, Subbéticas Moun tains Natural Park, 30S UG 81 49 U.T.M. coordinates, 1,000 m altitude) from October 1996 until the end of 1997, using twenty permanent pitfall traps each consisting of two concentric cylindric plastic pots (1,000 cc volume and 8 cm 0), using the out side of them as a container and the inside one as a recipient.Pitfall traps were mid-filled with a mixed solution of acetic acid as bait and ethanol (70°) as preservative, buried up to the top end, partially covered to avoid inundation and randomly distrib uted in the area considered.Systematic sampling was carried out at fortnightly intervals.
Anatomical observations related to the sex and age and to the reproductive condition of the females Age was determined by testing the softness of the integument and the extent of mandible wear, of the cephalic and thoracic bristles, and of the tibial spines and tarsal claws.Adults were placed in the following age classes: callows (or immature, with a very soft integument and unworn structures), young imagoes (with a hard integument but scarcely worn structures) and old imagoes (hard integument, and fairly worn structures, so that they were probably over one year old).In the case of females, the gonadal stage and potential fecundity were also checked.Using the same criteria as Cardenas (1994), after dissecting the females, they were classified as: -Teneral females (T), callow beetles, which ovaries are not yet developed.
-Pre-reproductive females (PRF), maturing females, with oo cytes clearly diferentiated but without mature ova or corpora lu tea.

The reproductive biology of C. dufouri
Laboratory rearing experiments were carried out in culture rooms, under different conditions of temperature, humidity and photoperiod in order to study their effect on the reproductive bi ology, oviposition patterns and developmental details of C. dufouri.
We started with ten reproductive groups or parentals (a male and a female in each) sampled in La Nava in October 1996.Pre vious anatomical studies referring to ovarian state indicated that C. dufouri had begun reproduction in this month.Each repro ductive group was kept in a cylindrical glass container (12 cm diameter, 14 cm height) with 5 cm thick moistened substrate of peat and a plentiful supply of maggots as food and arranged, in a shaded place adjacent to the laboratory.To indicate environ mental conditions we include the monthly temperature mean, relative humidity, monthly precipitation and approximate photo period (Fig. 2).
Eggs were kept at 18-20°C and in full-darkness.For larval rearing we followed the methodology proposed by van Dijk (1973) and Mols et al. (1981).Two groups of larvae were arranged in different conditions to determine the possible effect of temperature on the duration of development, and sur vival rate of immature stages.The first group was kept outdoors, while the second group was maintained under controlled condi tions (20 ± 1°C; » 85% R.H.; 12L : 12D).
The following parameters were recorded: Potential fecundity.For a specific time period, this was estab lished as the mean number of eggs per fertile dissected female; considering a fertile female to be one which had at least one ma ture ovum in ovaria.
Realised fecundity.This was defined as the mean number of eggs laid per female in a time period.

Statistics
Finally, for analyzing and comparing the results, the Mann Whitney non-parametric statistical test was applied (Zar, 1984).

Temporal activity pattern
The temporal activity of adult C. dufouri is plotted in Fig. 3, showing two maxima, the first between April-June and the second between October-December.The periods of peak activity are not homogeneous due, probably, to sampling effects.When the activity for males and females are considered in isolation, their respective curves run parallel to those of the summed data, although the males start moving and peak in activity sooner than females.The sex-ratio is favourable for females independently of the time of year.The low activity periods are associated with extreme climatic conditions (winter or summer), but some specimens can be found at these times.

Age-class distribution of C. dufouri
Following the criterion previously specified (see Mate rial and Methods section) anatomical studies enabled us to establish three age-classes as represented in Fig. 4. The majority of the members of the population corresponded to individuals belonging to the last generation.At the end of the winter (February-March), the majority of speci mens disappear (probably dying) and only a very small fraction of old imagoes survive until the next spring, and are found alongside the new generation.At the beginning of the spring, the first freshly emerged callow imagoes were caught.Emergence peaks in April but is spread out until July.Adults in different states of tegumentary hard ening and physiological maturation are found during the main activity period.

Temporal change of female reproductive state
For the course of the reproductive state of females see Fig. 5.The dissected females were teneral, immature, pre reproductive, gravid, females ending reproduction or spent females (see definitions in Material and Methods).In the annual cycle there is no repetition of any physio logical stage; this is typical for a univoltine species.
After the unfavourable summer, in mid September pre reproductive and gravid females were simultaneously found, which indicates that after aestivation the ovaria start differentiation and development.The oviposition pe riod begins at the end of September and continues until the middle of January when the females finishing repro duction disappear, and spent females are present in Febru ary and March.Later, most of them probably die and are replaced by teneral and immature females of the new gen eration.

The course of the oviposition
To estimate the reproductive potential the number of ripe eggs in the ovaria throughout the oviposition period was related to the eggs laid.On the basis of an observation of dissected females and the number of eggs laid in the rearing experiments, it was possible to evaluate temporal change in oviposition dur ing the entire breeding period.The number of ripe eggs in the ovaria was considered to be potential fecundity, while the number of eggs laid expressed realised fecundity.
Potential fecundity did not experience great fluctuation (Fig. 6).The last females with mature eggs were found at the end of December but in January the formation and ripening of eggs seem to cease.
In relation to realised fecundity, the fortnightly average values approximate those of potential fecundity and the length of both periods (when egg maturation and egg deposition occurs) is similar; nevertheless, a delay of ap proximately one month occurs between both processes.The oviposition period was quite long, taking in the whole of the autumn and the first days of winter.The av erage number of eggs per oviposition for each female ranged from 1.25 to 5.62 (Table 3) and the maximum number of eggs from one oviposition event was 12.
Nevertheless, when the regression between the values of both fecundity estimates is calculated, no significant correlation is found (r = -0.74;P = 0.26).On the other hand, if from the number of ripe eggs found in the ovar ian the corresponding number of eggs laid (estimated fe cundity) is calculated by the Gram equation (Gram, 1984) neither between potential fecundity/estimated fecundity nor realised fecundity/estimated fecundity significant  corelations are obtained (r = 0.12; P = 0.69 and r = 0.25; P = 0.68, respectively).There were important differences in the number of eggs laid by individual females (Fig. 7).Most of them had their egg-laying periods between November and Decem ber, right in the middle of the oviposition time established for C. dufouri in the previous paragraph.The total num ber of eggs laid per fertile female in the rearing experi ment (under outside conditions) ranged from 2 to 67.A low reproductive rate for C. dufouri might be expected for a Carabus species, compared with other Carabidae (Casale et al., 1982).

Breeding experiments: Biology of immature stages
As happens in most carabids, the development of C. dufouri goes through three larval instars and a preimaginal pupal phase.Duration and survival rates of each stage will be analyzed on the basis of environmental conditions (Table 1).
To improve egg-hatching and ensure a suficient number of eggs, the eggs were kept in full-darkness and at con stant temperature.For this reason, no available data exist about the incubation time and the survival rate for embry onic period under environmental conditions.Neverthe less, under controlled conditions the mean incubation time was around 12 days and the survival rate reached 69%.
Rearing of larvae under different environmental condi tions provided data on the duration and mortality of each stage (Table 1).There was high survival in all stages, es pecially in the second larval instar, whose survival rate exceeded 95%.The first and third larval instars had sur- vival rates of between 80 and 85%, while the preimaginal stage did not exceed 63%.
With regard to the duration of the different develop mental stadia, the first and second larval instars were similar («20 days).However, the third larval stage lasted for almost 50 days.The mean time for the preimaginal stage was again shorter lasting approximately 16 days.In summary, mean total developmental time was 104.6 days, and the total survival rate exceeded 40%.
To test the effect of temperature, a parallel culture was kept under controlled conditions (see Fig. 2) starting from a similar number of larvae in first instar, kept in equal substrate conditions and supplied with an equivalent amount of food.Substantial differences can be observed in relation to the duration and the survival rates of each instar (Table 1).In all stages the development was shortened and sur vival reduced (mean total developmental time: 55.34 days, total survival rate: 16.95%).To evaluate the signifi cance of the differences observed between the two experi mental groups the Mann-Whitney-Wilcoxon test was ap plied.The Z values and the respective significance levels are recorded in Table 2. Very significant statistical differ ences occur when the mean values of developmental time of each stage are compared, strongly suggesting that tem perature influences development rate.
Also, an additional fact is gleaned from the laboratory study when the duration of the developmental stages is re lated to the breeding period.In Fig. 8 (A and B) the course of the experiment (days) is related to the larval de velopment time (from LI-LIII) and duration of the prepu pal stage (L-III to Pupa).In both graphs a shortening at the end of the breeding period can be observed even when culture conditions are kept constant.This fact is shown if the regression of the duration of the different stadia is calculated over time (r = -0.43,P = 0.006 and r = -0.56,P = 0.001 respectively).
The same situation was recorded for outdoors (Fig. 8 C  and D), but a better regression and a higher slope are ob tained when the temperature also affects the duration of the developmental instars (r = -0.56,P < 0.0001 and r = -0.89,P < 0.0001 respectively).

DISCUSSION
In Temperate Zones the life-cycle of ground beetles is firstly determined by annual fluctuations in environmental (climatic) conditions and, consequently, they are likely to be univoltines.Our results show that C. dufouri has a sin gle generation each year and that its life-cycle belongs to autumn-breeders (sensu Thiele, 1977).The species shows a marked rhythm in activity and requires enough humidity and relatively low temperatures for oviposition and devel opment.In the South of the Iberian Peninsula, the sum mer is an unfavourable season for C. dufouri when it shows the lowest surface activity, adapting itself to cli matic conditions by taking refuge under trunks and stones probably in a state of gonadal dormancy (aestival para pause, Thiele, 1969).It is known that for species showing this developmental form, dormancy is terminated by changes in photoperiod from long day to short day.Thus, at the beginning of September the imagoes renew their activity, the males approximately two weeks earlier than the females.This delay is predictable when reproduction starts since the searching-activity anticipates oviposition.Whenever the activity of both sexes coincides, the sex ra tio is female biased.This last fact may be associated with the effect of the pitfall-traps on the aggregation of indi viduals (different for males than for females), but it may also be interpreted as a manifestation of a real numerical difference that tends to compensate for scant dispersal ability (due to the absence of wings) and the low repro ductive rate of this species, as will be commented on in the next paragraphs.
Our results for C. dufouri give values of 19.44 eggs/1/season and 6 ovipositions/1/season (a = 21.84 and a = 5.61 respectively), the average number of eggs/oviposition being 3.06 (a = 2.33).These data to gether with those mentioned in the previous sections and included in Table 3 confirm that the species has a low re productive rate but a high probability of developmental success (nearly 70% of the eggs hatched and, under out side conditions, 50% of them completed development).If the reproductive potential is estimated following the Gram (1984) method based on the ripe-eggs found in the ovaria between November-December (when centred on the oviposition, since not enough data are available for the total egg-laying period), it gave a similar value (18.04 eggs/ 2) although those deposited at the beginning or at the end of the reproductive season (October, January re spectively) are not considered and in any case they would have little influence on the outcome.
The lack of a linear correlation between the values of the potential oviposition, the real oviposition and those calculated from Gram's equation is explicable because the value of n (number of ripe-eggs found or the number of eggs laid by the culture females) are insufficient in all cases; so it is not possible to obtain definitive conclusions in this regard.
The establishment of the age classes led us to affirm that the majority of the specimens active in autumn had emerged the previous spring and belonged to the new generation.The oviposition activity peaks in mid autumn, when females in maturation state and gravid fe males are the only components of the female population.In December gravid females and females ending repro duction are found together in pitfall-traps and the first spent females are trapped in January.The progressive substitution of the different kinds of reproductive females shows the evolution of the oviposition phases.The sur face activity ceases sooner in males than in females; in December, few or no males are caught in traps.The new generation emerges in mid-spring, without go ing through an obligate diapause in any of the immature stages, as has been demonstrated by the cultures carried out under different experimental conditions.In both cul tures, full development was possible, but at constant tem perature the development rate was faster than in the outdoor culture, completing all the stadia in approxi mately half the time (55 days).The literature data indicate that humidity and temperature are the most limiting fac tors for carabid larval development.As the relative hu midity was similar in both cultures (70-80%) it seems clear that the temperature was the principal factor.On the other hand, temperature also exercised a negative influ ence, since in the laboratory the mortality rate was dou bled (41.66% of the first larval instar reached the imaginal stage under outside conditions, while only 16.95% did the same at constant temperature).Neverthe less, survival was similar for larval instars (66.67% out doors and 54.25% under controlled conditions), the pupal stage being clearly affected by temperature (62.5% as compared to 31.25%).High humidity and temperature in culture favoured the proliferation of acari that attacked the pupae, as the immobile stage is particularly vulnerable.Later it was verified that for pupae taken off the substrate and isolated in an aseptic medium (sterile sand) their probability of reaching the adult stage in creased and was similar to those pupae kept outdoors.In conclusion, temperature accelerates development, but does not decrease the viability of individuals.
In the field, the duration of the preimaginal stages (ap prox.100 days) is confined to the winter period in which the adults show scarce or no surface activity.In spring, the new generation emerges and is only coincident with a few specimens of over one year of age.Later, as the sea son passes, and depending on the local or climatic par ticularities, the imagoes will go into a state of inactivity, ending the life-cycle of C. dufouri, as is recorded in Fig. 9. Our experiments do not indicate which environmental factor induces this developmental pattern; a subsequent paper will be devoted to this topic.
In any case, these results are in contrast to what Hurka (1986) found for the Temperate zone autumn breeders, i.e. that it was obligatory to go through a temperature controlled larval dormancy in winter to complete their de velopment.According to the same author, habitat and its microclimatic factors limit the occurrence of certain types of annual reproduction rhythms but do not cause their evolution.Nevertheless, according to the findings of Paarmann (1979a, b) it is possible to establish clear evo lutionary relationships between certain types of annual cycle-life.It seems to have been well demonstrated that the annual rhythm of autumn-breeders (with gonad dor mancy during aestivation and temperature controlled lar val dormancy in winter, Paarmann, 1979a) is in some way connected to that of the winter breeders of the north of Africa (with gonad dormancy during aestivation and first indication of a temperature controlled larval dormancy in winter), and these cycles are the extreme (northern-most and southern-most) variants of one type of annual repro duction rhythm which perhaps finds its optimal expres sion in the Mediterranean region.In effect, the detailed study of the activity, in the different stadia, of C. dufouri shows an intermediate life-cycle type supporting Paarmann's assertions, because it basically corresponds to the type 4 of autumn breeders, but is also related to that of the winter-breeders due to the temporal delay in the period of maximum surface activity of adults and the larval devel opment free of obligatory interruptions.
Besides the cited temporal delay and the extent of the reproductive period, the results indicate that a shortening of the duration of developmental stages at the end of the breeding time is independent of the rearing conditions.This fact may be interpreted as a synchronization in the emergence of the new generation because the particular climatic conditions in the studied area (in late spring) has ten the completion of the life-cycle.In this sense, Paar mann (1979b) verified that the shortening, even the sup pression of developmental stages is one of adaptative strategies acquired by carabid beetles to survive in dry habitats and extreme aridity.
Finally, when the developmental cycle of C. dufouri is compared with those of other species closely related taxonomically, also belonging to the Mesocarabus subgenus, where the annual periodicity has been reported, i.e. C. lusitanicus and C. problematicus, differences are seen.The life cycle of C. lusitanicus, also in the South of the Ibe rian Peninsula (Cárdenas & Hidalgo, 1995), is similar to that of C. dufouri, but a slight temporal delay in oviposition time is observed in C. lusitanicus in respect to C. du fouri, as well as some asynchronisms in the duration of the preimaginal stages and, also, in the extent of the emer gence time of the new generation.These phenomenon may be adaptative responses to the particular microcli matic conditions inherent to the altitudinal differences be tween the respective natural areas of origin (Sierra Morena and Subbéticas Mountains).
The developmental cycle of C. problematicus in north ern England (Houston, 1981) is different from that of C. dufouri in southern Spain.Even though both species are autumn breeders (Larsson, 1939), the activity pattern in the field varies, with C. problematicus inactive in winter and C. dufouri in summer, probably as a consequence of the geographical location of their distribution areas.As a result, C. problematicus behaves as a larval hibernator while the adults of C. dufouri go through an aestivation period.Moreover, C. problematicus has a biennial life cy cle, but the majority of C. dufouri individuals live for a year in our research area.
On the other hand, in Holland (van der Drift, 1951) and in the south of England (Greenslade, 1965), adults of C. problematicus emerge in the spring, aestivate during the summer and breed in the autumn.This is similar to the cycle observed for C. dufouri in the Iberian Peninsula.
In summary, the data referred above support the state ment that carabids living in the Temperate zone show cli matic adaptations which affect the duration and the strategy of each developmental stage and alter their life cycles independently of phylogenetic relationships.

Fig. 1 .
Fig. 1.Climatogram of the research area for the sampling period (October 1996 to December 1997).

Fig. 2 .
Fig. 2. Monthly data of mean temperature, mean relative hu midity, precipitation and photoperiod for the outdoor culture.

Fig. 8 .
Fig. 8. Linear regression of the duration of the developmental stages over the course of the breeding experiments: A) For the two first larval stages kept under controlled conditions.B) For prepupal stage kept under controlled conditions.C) For the two first lar val stages kept under outside conditions.D) For prepupal stage kept under outside conditions.

Table 3 .
Mean, standard deviation, maximum and minimum number of eggs per oviposition for each fertile female from culture.