IMPACT OF WEATHER CONDITIONS ON SEASONAL DEVELOPMENT, POPULATION STRUCTURE AND REPRODUCTIVE SUCCESS ON DACTYLORHIZA TRAUNSTEINERI (ORCHIDACEAE) IN THE KOMI REPUBLIC (RUSSIA)

Due to specific biological features, high ornamental value and low resistance to anthropogenic factors, Orchidaceae species belong to the most vulnerable plants. To successfully protect their populations, comprehensive investigations of orchid biology and ecology are necessary. Long-term population studies are the most valuable. The paper presents data of population studies of Dactylorhiza traunsteineri, an orchid species listed in the Red Data Book of the Russian Federation. The field investigations have been conducted in the Komi Republic where the orchid species is located at the northeastern limit of its range. The seasonal development of D. traunsteineri lasts from May to August. We found the weather factors (air temperature, precipitations) impact features of small and big life cycles of this orchid species. The size of plant individuals is influenced by weather conditions of both the current and previous vegetative season. The number of generative (flowering) individuals per population had a positive correlation with the air temperature and humidity in August of the previous vegetative season. The fruit set of D. traunsteineri is high (50.4%). This parameter is negatively correlated with the air temperature at the flowering period, while it is positively correlated with precipitation values. The seed number per capsule (4090 seeds in average) was higher than it is known for other Dactylorhiza species in the Komi Republic. The real seed production is associated with the moisture content level during the vegetative season. The seed production of D. traunsteineri was high, from 88 000 to 199 000 seeds per 1 m2 in different study years. The presence of juvenile individuals (3.5–9.4%) over all study years indicates a successful seed reproduction in this population. This parameter was positively correlated with precipitation, air temperature in August, and seed production at the previous vegetative season.


Introduction
Orchidaceae is one of the largest families among flowering plants (Christenhusz & Byng, 2016). At the same time, the majority of orchids are rare, endemic, and threatened (Cribb et al., 2003). Orchid vulnerability is mainly associated with biology features, such as unique «orchid mycorrhiza», low competitiveness, and high specialisation of pollination (Shefferson et al., 2019). To successfully protect orchid populations in nature under conditions of anthropogenic pressure, comprehensive studies of ecology and biology of these vulnerable plants are necessary (Fay, 2018). Despite of a high interest in these plants, numerous issues of orchid biology are still unsolved. This is partially explained by the short-term character of most of the studies worldwide. But such results do not allow correctly assessing the current status of orchid populations and understanding their future survival prospects (Vakhrameeva et al., 2011). In this relation, the long-term studies of orchid populations at the fixed study plots are the most valuable. The long-term demographic investigations provide the inestimable data to understand how and why natural orchid populations are changed in terms of their size and structure (Hutchings, 2010). The need of monitoring populations of threatened orchid species was indicated also by Wraith et al. (2020) in a study devoted to an analyse of gaps and priorities in investigation and conservation of these vulnerable plants.
In the present paper, we have studied the impact of weather factors (air temperature and precipitation) on the reproductive success and population structure of a threatened orchid, Dactylorhiza traunsteineri (Saut. ex Rchb.) Soó s.l., at the northeastern limit of its range within the Komi Republic (Russia). Dactylorhiza traunsteineri is listed in the Red Data Book of the Russian Federation (2008) with category 3 (rare species). However, in Russia, this species has not been actually investigated (Vakhrameeva et al., 2014), except for the study of Blinova & Uotila (2012) in the Murmansk region. There are a few publications conducted in Norway (Øien & Pedersen, 2003;Øien et al., 2008;Sletvold et al., 2010aSletvold et al., ,b, 2013Sletvold & Agren, 2011, 2015 and devoted to the study of biology features of a closely related species, Dactylorhiza lapponica (Laest. ex Hartm.) Soó, which is sometimes considered as conspecific to D. traunsteineri s.l.

Measurements of morphometric parameters of individuals
Of 30 individuals in the field, we measured annually the plant height, inflorescence length, number and size of leaves, number of flowers and fruits (capsules). To reveal the size of flower parts in a certain individual, two flowers were mounted on cardboard from the middle part of the inflorescence using a transparent adhesive tape. Under laboratory conditions, electronic images of flowers were obtained using a scanner. Subsequently, we used them to measure their parts (lip, spur, petals) using Gimp 2.8 software.
To study seeds, we collected annually from 10 plants the fruits with matured seeds from the middle part of the inflorescence before they begin to open. Seeds were analysed using an MSP-2 light microscope (4.5 × magnification) with a TS-500 digital video camera (LOMO, Russia). The measurements were performed on digital photographs using ToupView software (ToupTek, China). We estimated the average length and width of each seed and embryo and their volume (Arditti et al., 1979;Healey et al., 1980) for 40-50 seeds from samples of each study year. To determine the seed quality, we used a mixture of seeds from capsules selected from different plants within the studied population (at least 600 seeds). The seeds were examined under the MSP-2 light microscope, marking seeds with embryo and seeds without normally developed embryo. The calculation of the seed number per capsule was carried out using the method we developed to analyse digital images of orchid seeds using the ImageJ software package (Kirillova & Kirillov, 2015, 2017a.

Estimation of population structure
To map the population, estimate its area and number of orchid individuals, we used the original method we developed on the basis of the joint use of aerial photography material (obtained using unmanned aerial vehicle DJI Phantom 2 Vision+) and results of the field study on the orchid population.
In 2015, we established the fixed study plot (15 m 2 ), where we annually indicated the location of all D. traunsteineri individuals with estimation of their ontogenetic stages. This allowed monitoring the development of each individual from appearance to dying. The determination of ontogenetic stages was conducted using widely accepted approaches taking into account specific features of orchids (Blinova, 1998; Vakhrameeva, 2000). We determined the following ontogenetic stages: juvenile (plants with one leaf of middle formation with 2-4 veins), immature (plants with 1-2 leaves of middle formation with 6-8 veins), mature vegetative (2-3 leaves with 8-12 veins) and generative (flowering plants). We did not determine senile plants as in nature these are observed rarely because many orchids die off after the last flowering (Vakhrameeva, 2000). During 2015-2019, we estimated the number of D. traunsteineri individuals on the study plot.

Collection and calculation of weather parameters
Data on the air temperature and precipitation amount on the study site in 2013-2019 are presented in Table 1. Characteristics of the vegetative season over the study period are described in Table  2. Values of the air temperature at the surface layer were measured using Thermochron DS1921G loggers, established near the orchid population at the height of 2 m above ground level. The frequency of temperature measurements was every 180 minutes. The measurement time was associated with standard synoptic periods. The precipitation amount was estimated using the open access source of «Array of urgent data on the main meteorological parameters at the stations of Russia» (http://aisori-m. meteo.ru). We used data from meteostation «Syktyvkar» (WMO ID 23804), located at 12 km of the study site (61.67720858° N, 50.78470815° E). On the basis of the obtained meteorological data, we calculated the Selyaninov Hydrothermal Coefficient (HTC) using the following formula: , where R -sum of precipitation amount (mm) over the period with temperature values > 10°C; ∑ t -sum of temperature values (°C) over the period with temperature values > 10°C.

Statistical analysis
Preliminary processing and analysis of data were conducted using Microsoft Office Excel 2010. Statistical analysis was processed using R (v.3.4.2) (The R foundation). In the main text and tables, we have indicated the arithmetic mean (M) and standard deviation (SD).
We used the Shapiro-Wilk W-test to check the samples of the values of the morphometric parameters of plant individuals, fruits and seeds for the normal distribution. To compare the samples, we used parametric methods (Student's t-test for sam-ples with normal distribution) and non-parametric methods (Wilcoxon-Mann-Whitney test for data with deviations from the normal distribution).

Seasonal development of plant individuals
In the Komi Republic, the average duration of the vegetative season (with the average daily air temperature values > 5°C) was 149 days (with variation of 142-170 days, between 26 April and 11 October) over the study period. The seasonal development of D. traunsteineri lasts from May to August (Fig. 3), while aboveground shoots die out in September. The vegetation begins often in mid-May. But sometimes this period shifts being related to different dates of vegetation period starting. For example, the earlier (8 April) beginning of the vegetation period in 2016 led to the fact that plants appeared earlier -in early May. The late dates of beginning of the vegetation period in 2017-2018 (11-18 May) shifted dates of D. traunsteineri's vegetation period to early June. The flowering begins at mid-June, its peak -in the third -fourth weeks of June. However, the flowering stage could also be shifted depending on the weather conditions. In 2016, plants flowered earlier, and the flowering peak was at mid-June. This was related to a very warm early summer (average air temperature of min-June was 17.8°C with a norm of 14°C) in this vegetative season. Vice versa, in 2017 and 2018, the flowering stage was shifted to early July. In August, the fruits ripen, darken, crack, and seeds begin to spill out. At this time, the renewal bud becomes noticeable. From this bud a young shoot will be developed the next year. By autumn, both young vegetative and young generative organs of future plant will be formed in the renewal bud. In such form, the renewal bud is wintered, while in spring, a new shoot will be developed.

Morphometric parameters of plant individuals
The morphometric parameter study for D. traunsteineri generative individuals demonstrated that the mean plant height was 21.30 ± 4.60 cm on the study site. Each plant has an average of three leaves. The bottom leaf is 7.29 ± 1.86 cm long and 1.31 ± 0.27 cm wide, the second leaf from the bottom is 9.0 ± 1.49 cm long and 1.26 ± 0.29 cm wide. The inflorescence is 5.04 ± 1.19 cm long, with 14.9 ± 4.90 (from 6 to 29) flowers. The petal length is 10.0 ± 1.03 mm. The flower lip is 8.91 ± 0.89 mm long and 10.62 ± 1.35 mm wide.  Morphometric parameters of generative individuals of D. traunsteineri at different years are presented in Table 3. The height of the plants varied from 15.9 cm to 25.7 cm. The highest value was found in 2015, while the lowest value in 2016. The vegetative season of 2016 was the warmest over the whole study period (Table 2). We found a positive significant correlation (r = 0.9) between plant height and inflorescence length. Both these parameters are negatively correlated with the precipitation amount in early May of the current year, and they are positively correlated with the air temperature values in the first half of August of the previous vegetative season (Table 4). The leaf size had maximal values in 2015, while a minimal size of leaves was found in 2018 (Table 3). The size of the bottom leaf was positively correlated with the air temperature values early June of the current vegetative season, while the leaf length had a positive correlation with the precipitation amount in August of the previous vegetative season (Table 4).     Depending on the study year, the number of flowers varied from 11.3 (in 2016) to 17.2 (in 2019). The flower size varied over the years, too (Table 3). This parameter was correlated positively with the sum of active temperature values (> 10°C) in the whole vegetative season (r = 0.9) and with the air temperature in July of the current year (Table 4). The largest flowers were found in the warmest year, 2016, when we noted the maximal sum of active temperature values (1548.1°C), while minimal values were noted in the coldest vegetative season, 2019, (946.9°C) ( Table 2). The flower size was negatively significantly correlated with the number of flowers (r = -0.9).

Structure and dynamics of population
In the studied population, the total D. traunsteineri abundance was 23 722 plants at an area of 11 302 m 2 . In the study plot, the number of individuals increased slightly in 2017 ( Fig. 4) after the warmest and sufficiently humid year of 2016. This parameter sharply decreased in 2019, i.e. after the driest vegetative season of 2018, when we found the minimal value of HTC (Table 2). The number of individuals was positively correlated with the precipitation amount in late June and the second decade of September of the previous vegetative season (Table 4), while this parameter was negatively correlated with HTC of the previous vegetative season (r = -0.9).
Over all study years, the age structure of the D. traunsteineri population contained individuals of all ontogenetic stages with a maximum of mature vegetative individuals (Fig. 4). The highest number of generative individuals was found in 2015 and 2017, while the lowest number of flowering plants was registered in 2016. The number of generative individuals was positively correlated with the air temperature and precipitation amount in August of the previous vegetative season (Table 4), as well as with the precipitation amount over the period with an air temperature > 10°C in the previous vegetative season (r = 0.9).
In 2015, 7.7% of the flowering plants were damaged. This was probably influenced by frosts in October of the previous year (air temperature of the second decade of October was -5°C). In this period, frosts are harmful to the future generative plants because in autumn flower primordia have already been formed in buds, from which the next year inflorescences will develop.
The observation of the mapped orchids showed that 57% of the generative individuals flowered only one time, 37% of the plants -during only two subsequent years, and only 6% individuals flowered over three years. The next year after flowering, most of the individuals (64%) turned into the temporarily non-blooming stage, which lasted 1-2 years. Sometimes after a year spent in a mature vegetative stage, the plants bloomed again. In addition, 36% of the generative plants died after flowering.

Reproductive biology
Over six study years, the average fruit set of D. traunsteineri was 50.4%. Its highest values (more than 62%) were found in 2017 and 2019, while the lowest values (37.5-39.0%) were recorded in 2015 and 2018 (Fig. 5). The number of the formed fruits correlated with the weather conditions in June (the time of D. traunsteineri flowering). This parameter was negatively correlated with the air temperature and positively correlated with the precipitation amount over this period. The fruit set does not depend on the number of flowers per inflorescence or flower size.
The seeds of D. traunsteineri are light brown. Mature seeds are composed of a transparent seed coat and an undifferentiated germ (Fig. 6). We found some singular seeds with two embryos. In the studied population, the average seed length was 0.71 ± 0.10 (0.41-0.93) mm, seed width -0.20 ± 0.04 (0.13-0.32) mm. The seed index was 3.6. The embryo size was 0.22 ± 0.03 mm long and 0.14 ± 0.02 mm wide. On average, 67.5% of a seed is occupied by an empty airspace.
The seed volume was positively correlated with the embryo volume (r = 0.9) and the fruit set (r = 0.8). In the years characterised by a low fruit set (2014,2015,2018), we found the lowest size of seeds (Table 3). The embryo volume was positively correlated with the air temperature in early July, i.e. a period when fruits are being formed.
The fruit of D. traunsteineri is a capsule. In the Komi Republic, its average size is 12.0 ± 1.28 × 4.23 ± 0.65 mm. The highest size of fruits was registered in 2018 and 2019, while minimal values were found in 2015 (Table 3). One capsule of D. traunsteineri contains on average 4090 ± 145 seeds (minimum -2352 seeds, maximum -5197 seeds). The highest average seed number (4300) was found in 2016 and 2019, the lowest average value (3700) -in 2015 (Table 5). We found a positive correlation between the seed number and the fruit width (r = 0.8).
Some proportion of seeds does not contain a normally developed embryo (Fig. 6b, Table 3). These seeds have a lower size -0.42 × 0.10 mm, i.e. they are almost two-fold smaller than seeds with an embryo. The proportion of seeds without an embryo was positively correlated with the precipitation amount in June. The highest number of seeds without an embryo was found in 2014, when we registered a maximal precipitation amount in mid-June -63 mm, that is two-fold higher than the norm (26 mm).
In different years, the conditionally potential seed production of generative plants (seed production with 100% pollination efficiency) varied from 48 800 seeds to 75 300 seeds per individual. The conditionally real seed production (where pollination efficiency is taken into account) varied from 22 000 seeds to 46 800 seeds per plant (Table 5). Over five study years, the average real seed production was 29 487 seeds. The maximum average value (42 884 seeds) was found in 2019, while the minimal average value (20 960 seeds) was registered in 2018. The real seed production was positively correlated with HTC (r = 0.9), which indicates the level of water content in an area, as well as the precipitation amount in late June.
We calculated the seed yield by multiplication the real seed production on the average number of generative plants per 1 m 2 . The number of generative plants per 1 m 2 varied in different study years with a variation from 2.6 individuals per 1 m 2 (in 2016) to 5.4 individuals per 1 m 2 (in 2017). The maximal seed yield was found in 2017 with 199 100 seeds per 1 m 2 , while a minimal value (61 600 seeds per 1 m 2 ) -in 2016. The seed yield was positively significantly correlated with the sum of active temperature values of the previous vegetative season (r = 0.9).   Although D. traunsteineri individuals form a large number of seeds, only a small part of them could germinate. Over all study years, the proportion of young plants was 3.5-9.4%. Its maximal value was found in 2017 after the warmest and sufficiently humid season of 2016, when the highest number of plants in the population was observed. The minimal number of juvenile individuals was noted in 2019 after the driest season of 2018 (HTC had a minimal value, see Table 2). The number of young individuals was positively significantly correlated with the precipitation amount and the air temperature in August of the previous vegetative season (Table 4), as well as with the seed yield in the previous year (r = 0.9).

Discussion
Numerous investigations covering different organisms, demonstrate that changes in parameters of their life activity are associated with climate changes, such as temperature and precipitation. It was repeatedly reported that climatic factors affect the size, flowering likelihood, reproductive success, population dynamics, and survival of orchids (Blinova, 2008;Inghe & Tamm, 1988;Wells & Cox, 1991;Wells et al., 1998;Carey & Farrell, 2002;Pfeifer et al., 2006;Jacquemyn et al., 2009). Our results have also demonstrated that weather conditions influence features of small and big life cycles of D. traunsteineri at the northeastern limit of its range. So, the plant size is affected by weather conditions of both current and previous vegetative seasons. A warm beginning of the vegetative season positively impacts the leaf width. The air temperature in early July is important for forming the large flowers, like Øien & Pedersen (2003) showed that the aboveground biomass of the flowering D. traunsteineri individuals increases during the whole flowering period. The air temperature and the precipitation amount in August of the previous vegetative season influence positively the height of plants and the inflorescence length. Moreover, the precipitation in late August of the previous year is positively correlated with the size of future leaves of the orchid individuals. This could be explained by the fact that in this period, the organs of the future shoots are actively forming in the bud, which requires a stock of nutrients. During this period, the biomass of new tubers actively increases even after the photosynthetic assimilation ceases. This occurs due to mycotrophic activity and re-location of nutrients from aboveground shoots (Øien & Pedersen, 2003), while positive temperature and sufficient humidity favourably affect the activity of the mycotrophic component. We noted the negative correlation between the number of flowers and their size. Thus, the plant spends resources either on the formation of a lower number of the larger flowers or a higher number of smaller flowers.
The proportion of generative individuals changed two times over the study period (from 19% to 37%). The considerable inter-annual fluctuations in the number of flowering orchid individuals has been demonstrated also by other studies (Tamm, 1972;Øien & Moen, 2002;Jacquemyn et al., 2007;Pfeifer et al., 2006;Hutchings, 2010;Van der Meer et al., 2016). The number of flowering individuals was positively correlated with both the temperature and precipitation in August of the previous vegetative season. In Central Norway, Øien & Moen (2002) also found a positive significant correlation between the density of generative D. traunsteineri individuals and the temperature of the previous vegetative season. This is associated with the fact that the flower primordia of this species are being formed a year before flowering (Øien & Pedersen, 2003).
The number of orchid individuals in the studied population (about 23 700 plants) is high for this species. So, in Central Norway, D. traunsteineri populations contain 400-500 flowering plants (Øien & Moen, 2002;Øien et al., 2008;Sletvold et al., 2010b), compared to about 50 individuals in the Murmansk region of Russia (Blinova & Uotila, 2012). A high moisture level in the current vegetative season negatively affects the size of the examined population. A similar pattern was noted for another orchid species, Neottia ovata (L.) Bluff & Fingerh., in the Murmansk region (Blinova, 2008). However, a warm and relatively humid previous vegetative season positively affects the number of individuals in a population.
The fruit set of D. traunsteineri is relatively high -50.4%. In Central Norway and the Murmansk region (Russia), this parameter value does not exceed 30% (Sletvold et al., 2010a;Blinova & Uotila, 2012). In the Moscow region (Russia), the fruit set is 10-15% with an increase up to 60-65% in the most favourable years (Vakhrameeva et al., 2014). The number of formed fruits depends on the weather conditions during the flowering stage. At the same time, hot and dry weather negatively influences the fruit set of D. traunsteineri. This is probably explained by the fact that Bombus species are pollinators of D. traunsteineri (Sletvold et al., 2010b). These insects can fly already at 10°C and pollinate flowers even in a cold and rainy season (Jakubska-Busse & Kadej, 2011), while their activity is decreased at a temperature > 23°C (Demidova, 2012). In addition, we repeatedly observed ants on D. traunsteineri flowers in the studied population. Perhaps, these insects also play a certain role in D. traunsteineri's pollination, as ants were noted as effective pollinators for some plant (including orchids) species (Rostás & Tautz, 2010;Schiestl & Glaser, 2012).
The fruits of D. traunsteineri contain a large number of small seeds. The seed number was positively correlated with the fruit width, the volume of seeds and embryos. On average, each D. traunsteineri capsule contains 4090 seeds. This value is higher than it was found for other Dactylorhiza species in the Komi Republic. For example, a Dactylorhiza fuchsii (Druce) Soó capsule contains on average of 2597 seeds (Kirillova & Kirillov, 2013), a D. maculata (L.) Soó fruit 2835 seeds (Kirillova & Kirillov, 2017b). In the Murmansk region (Russia), only 1200 D. traunsteineri seeds per capsule were noted (Blinova & Uotila, 2012). The real seed production was positively correlated with the water content of the whole vegetative season and the precipitation amount during flowering of the plants.
The result of a successful reproductive process is the number of emerged and established young plants. Over all study years, this parameter was on average 5.8%. A low number of juvenile individuals in D. traunsteineri populations was also noted by Blinova & Uotila (2012). Øien et al. (2008) demonstrated that seeds of this species are short-living (less than one year). They do not form a seedbank in the soil, and the natural germination is very low, only 11%. Therefore, seeds die, if the conditions are not favourable for germination. In addition, the contact of a seed with the mycelium of a compatible fungal symbiont is necessary for a successful development of the orchid seedling. The number of young D. traunsteineri individuals was positively correlated with the seed yield of the previous year, as well as with the precipitation amount and the August temperature of the previous vegetative season. The importance of moisture at the time of seed dispersal to the ground was also indicated for other orchid species (Kulikov & Filippov, 2001;Scott & Carey, 2002).