Seagrasses and the valuable ecosystem services they provide are threatened world-wide by impacts of human activity. Numerous revegetation efforts have attempted to restore seagrasses. Most restoration programmes use plants collected from the field because of limited seed availability, low seedling survival and difficulty in culturing plants. However, this practice risks damage to donor populations and has the potential to reduce genetic diversity, which may counteract the desired effects of restoration.
A novel aquaculture system for producing plants (mother plants and cuttings) from a limited number of seeds was tested using Cymodocea nodosa as model species. The ability of transplanted cuttings to survive and grow in the natural habitat was also evaluated.
Seed germination was high (48%) compared with field conditions, and most seedlings produced mother plants with up to 7·8 m of rhizome and 300 shoots within 4 years in culture. All cuttings from mother plants regenerated new plants. Up to 100 transplants were produced from two seeds, and most (85%) of them survived and initiated the colonization of substrate, 1 year after planting.
Synthesis and applications. This study provides a robust protocol that can reduce plant and/or seed collection pressure on donor populations and produce a high number of transplants which show lower mortality rates during the early transplantation phases. This method can also help to preserve genetic diversity in restored populations, which should be one of the major goals of ecological restoration. This novel tool can be applied to other seagrass species with low or unpredictable reproductive success, therefore the development of nurseries should be incorporated in future restoration programmes. This is currently the only sustainable methodology to produce material for transplanting programmes for protected species.
Seagrasses are clonal marine angiosperms which have experienced serious declines throughout the world because of anthropogenic influences (Duarte 2002; Green & Short 2003; Waycott et al. 2009; Short et al. 2011). Concerns over the maintenance of ecosystem functions and services provided by seagrasses on the coastal marine environment have stimulated major efforts to prevent further losses and restore degraded habitats (Hemminga & Duarte 2000; Duarte 2002; Orth et al. 2006a,b). As seagrasses may take decades or centuries to recover from disturbance over large areas (Walker et al. 2006; Uhrin, Kenworthy & Fonseca 2011), the active introduction of vegetation may be required to restore degraded habitats.
Most restoration efforts attempted world-wide have involved the translocation (transplantation) of vegetative material (adult plant cuttings) from healthy populations (donor populations) to restoration sites (Fonseca, Kenworthy & Thayer 1998; van Katwijk et al. 2009; Paling et al. 2009). However, this practice has recently received substantial criticism, contending that the removal of a large amount of plant material may damage donor populations, thereby counteracting the desired effects of the restoration action (Duarte 2002; Balestri, Vallerini & Lardicci 2011). A growing number of studies indicate that the use of sexual material (seeds and seedlings) represents an effective and more ecologically sustainable alternative method for restoration, providing the opportunity for preserving genetic variability in restored populations while reducing the impact on donor meadows (Balestri, Piazzi & Cinelli 1998; Seddon 2004; Orth et al. 2006a; Marion & Orth 2010; Tanner & Parham 2010; Zarranz et al. 2010; Orth et al. 2012; Kendrick et al. 2012; Reynolds et al. 2012). However, seed availability and vulnerability of seedlings to environmental stresses during initial phases of establishment may limit the applicability of this approach on a large scale. Restoration actions typically need a quite large supply of seeds (hundreds to millions of seeds) to compensate for the high losses and the low rate of colonization by seedlings (Golden et al. 2010; Marion & Orth 2010). For seagrasses such as Posidonia and Cymodocea, it may be difficult to collect a sufficient number of seeds because of infrequent flowering or low seed output (Duarte & Sand-Jensen 1990; Piazzi, Balestri & Cinelli 2000; Balestri & Vallerini 2003). In other seagrasses, such as Zostera, large quantities of seeds may be available annually (Orth et al. 2006a,b). However, excessive seed removal may affect the structure and dynamics of donor populations. This potential risk is generally neglected by managers, but should be weighed against the potential restoration benefits (Orth et al. 2006a,b; Marion & Orth 2010). To improve the chance of successful establishment in restoration actions, efforts have recently focused on the development of cultivation systems that allow large number of seedlings to be grown for several months prior to transplantation, in order to produce seedlings of a sufficient size to withstand adverse field conditions (Balestri, Piazzi & Cinelli 1998; Seddon 2004; Tanner & Parham 2010; Zarranz et al. 2010). Although effective, these systems do not solve the problem of seed supply and low seedling colonization rate. Thus, it is important to develop novel restoration strategies based on a more efficient use of limited seed supplies.
In this study, we described a novel procedure for producing a large amount of planting material for restoration in an aquaculture setting from seeds collected with a minimum impact on donor populations. We chose Cymodocea nodosa (Ucria) Ascherson as a model. This dioecious fast-growing species is common along the Mediterranean and Atlantic coasts (Green & Short 2003), but it strongly declined during the 20th century and for this reason is protected by legislative measures (Boudouresque et al. 2009). Attempts to restore meadows by transplantation have occurred recently (Zarranz et al. 2010). In a long-term production experiment, we evaluated the production of plants (mother plants) which can be generated from seeds, and in two propagation experiments, we tested whether further propagation and regeneration of new plants (stock plants) are possible by cutting fragments from mother plants. Specifically, we examined (i) seed germination, seedling survival and growth over time and the number of years plants can be kept in culture before saturating growth, (ii) the type of growing container (rectangular or round) most suitable for long-term permanence, as the form of C. nodosa patches changes from elongated to circular once they exceed 4–5 years (Sintes et al. 2005), (iii) the relationships between genet age, patch size and onset of sexual reproduction and the performance and investment in reproduction of male and female genets, (iv) the effect of season (spring vs. fall) and morphological fragment type (apical or nonapical) and variation between genets on the propagation potential of cutting fragments from mother plants. Finally, a field experiment was conducted to determine the transplantation success of cutting fragments from stock plants.
Material and methods
Cymodocea nodosa shows a clear unimodal annual growth cycle in the Mediterranean Sea, with a peak during June and July, and a cessation of rhizome growth from October to January (Caye & Meinesz 1985). The flowering frequency and the seed production rate vary among localities (Caye & Meinesz 1985; Terrados 1993). Seeds mature in spring–summer, remain dormant through the fall–winter and germinate in the following spring. Germination rate in nature is relatively low (50%), and ungerminated seeds may be incorporated in a seed bank (Terrados 1993). Most established seedlings (80–90%) do not survive to their first year because of nutrient limitation and competition with neighbouring seedlings in established meadows (Duarte & Sand-Jensen 1996; Balestri, Vallerini & Lardicci 2010), so the density of patch-forming seedlings in meadows is usually low, approximately 1 per 207 m2 (Vidondo et al. 1997).
Seeds of C. nodosa were collected from 16 randomly chosen areas (six seeds per area) spaced at least 30 m apart within a meadow near Livorno (north-western Mediterranean, Italy, 43° 30′ N, 10° 19′ E) at a depth of ca. 1 m, in August 2004. Seeds were transported to an aquaculture facility (Mariculture S.r.l. of Rosignano Solvay, Italy). The cultivation system consisted of four outdoor fibreglass tanks (5000 L) connected to a system that pumped natural filtered seawater, equipped following a protocol under Italy patent application (PI/2005/A/000092). Seawater temperature in tanks ranged from 11 °C (winter) to 30 °C (summer) during the study period. Soon after collection, seeds were sown 2–3 cm deep in sand collected from the donor meadow in five plastic rectangular containers (45 × 15 × 16 cm). There were 15 seeds per container. In May 2005, the number of germinated seeds was counted. Emerged seedlings were removed from the substrates (Fig. 1), selected for uniformity of size (1 leaf, ca. 2 cm leaf length) and transplanted individually in containers filled with a mixture (80% v/v) of beach sand (washed with seawater) and volcanic stone (Agrigallo, Torino, Italy). Two container types, rectangular (45 × 15 × 14 cm) and round pots (30 cm diameter, 14 cm height), were tested. There were six rectangular and six round containers, 10 cm apart, in each tank; providing 24 seedlings in total. They were placed at a depth of ca. 1 m and rotated within each tank over the study period. As two seedlings died during the first year, the number of containers per typology was reduced to ten. A controlled-release fertiliser, Pluscote (R) Garden (16% N, 8% P, 16% K), was applied bimonthly at a rate of 1·5 g l−1 of substrate (Balestri, Vallerini & Lardicci 2010) during the growing season over the study period. Survival and number of living shoots per plant were recorded in August of each year; morphological variables (leaf length, number of standing leaves per shoot and maximum shoot height) were measured on 10 randomly chosen shoots per plant. The proportion of flowering plants per container type, their sex and number of reproductive shoots per genet were also determined. Sex ratios were calculated as the proportion of male genets to sum of males and females. The number of shoots of flowering and nonflowering plants at the first flowering was compared to determine whether there was a threshold size for reproductive maturity. In addition, the flowering ratio of the shoots of males and females (number of flowered shoots/total number of shoots, expressed as a percentage) was calculated to investigate whether equal sex allocation occurred in sexual reproduction. The maximum reproductive potential (seeds m−2) was determined based on the number of female flowers per plant.
Finally, two randomly selected plants were extracted from the substrate in August 2009, and two additional plants were extracted at the end of the study (August 2011) for morphological measurements (shoot number, rhizome length, number and length of branches and length of rhizome network). Maximum shoot height and number of leaves per shoot were also measured on 10 randomly chosen shoots per plant. Flowered plants were excluded from this destructive investigation. To avoid possible confounding because of different growth conditions, plants from the same tank and container type (rectangular type) were measured.
Two experiments with a subset of randomly selected mother plants from the production experiment were conducted before the onset of reproduction. To examine the effect of season on propagation success, cuttings were excised at the start of the growing season (April 2009, spring experiment) and at the end of the growing season (late August 2009, fall experiment) from 4-year-old plants (Fig. 1); two plants grown in round containers and two plants grown in rectangular containers were used in each experiment. At each date, four horizontal rhizomes were randomly selected from each plant. Rhizomes were cut into unbranched fragments 12 cm in length with an average of 4·4 ± 0·45 shoots (mean ± SE) and planted in round containers (54cm diameter, 12 cm height) filled with a mixture of beach sand and volcanic stone (80% v/v). In the spring experiment, 16 of the fragments obtained were planted in four containers (experimental units); each container contained four fragments. Fragments were held for approximately 5 months, until the end of the first growing season (September 2009). In the fall experiment, 100 fragments were obtained from mother plants. Fragments were categorized as with and without terminal apical meristems to distinguish between morphological fragment type and genets. Studies have shown that apical dominance prevents a node along the horizontal rhizome of C. nodosa from generating primary lateral branches (Terrados, Duarte & Kenworthy 1997). Fragments with an apical meristem were excised from the youngest portion of horizontal rhizomes, while fragments without an apical meristem were obtained from the remaining portions of rhizomes. A subset of fragments (10) per morphological type and genet was used. Fragments were planted separately in containers as described above, such that there were two experimental units per fragment type and genet, each containing five fragments, 40 fragments in total. Fragments were kept for approximately 10 months, until an appreciable size increase was observed (June 2010). Slow-release fertilizer (Pluscote) was supplied at planting at the same rate as for mother plant culture. At the end of the experiments, the propagation potential was calculated as the percentage of fragments that produced at least a new shoot, and the total number of living shoots and branches, total rhizome length, branch length and maximum shoot length of newly generated plants were recorded.
Stock plants regenerated from mother plants in the propagation experiment were cut into fragments (Fig. 1). Fragments (32) of uniform size (approximately 7 cm in rhizome length, at least two shoots and the terminal apical meristem) were excised from the lateral branches in June 2010 and immediately transplanted at two replicate sites, ten metres apart, into a shallow coastal area (Rosignano Solvay), at a depth of 0·2–0·5 m, close to a C. nodosa meadow. The substrate consisted in sand and pebbles. Transplants were anchored to the substrate using metal pins (4 cm in length) that were removed soon after rooting. There were 16 transplants per site. Transplant survival and shoot production were monitored every month during the first growing season and then at the end of the experiment (May 2011). Maximum shoot height, total length of rhizome and number of branches per plant were recorded nondestructively at the end of the experiment.
In the production experiment, an equal number of mother plants survived to the end of study period in rectangular and round containers, thus no analysis was necessary to test for differences among treatments. Morphological variables were analysed using analysis of variance by permutation, permanova, according to mixed-model anova design with the orthogonal factors, container type (fixed, two levels) and tank (random, two levels). The interaction term was pooled with the residual to increase the analysis power. Because some plants were sampled during the study for observation and propagation, there were six plants per treatment. The number of flowering clones in each container type was analysed by a chi-square test for equality of proportions with a significance level of α = 0·05. A logistic regression was performed to test for possible relationships between flowering probability and plant size (i.e. number of shoots). In total, 14 plants (seven flowering and seven nonflowering) were examined. The sex ratio of flowering plants was tested against a null expectation of an even sex ratio (0·5) using an exact binomial test. A one-way anova was performed to determine whether there were differences in the number of vegetative shoots between sexes. A one-way anova was also performed on the flowering ratio to examine whether sexes invested equally in reproductive structures. Because of the low abundance of males, only three plants per sex were compared and data from the two container types (rectangular or round) were pooled.
In the propagation experiments, all fragments regenerated new plants, thus no analysis was necessary to test for differences among treatments. Morphological variables in the spring propagation experiment were analysed using permanova with experimental unit as random factor. permanova was also used to test for morphological differences among plants in the fall experiment according to a mixed-model anova design with the orthogonal factors fragment type (fixed, two levels) and genet (random, two levels), and experimental unit nested within fragment type and genet (random, two levels). The number of shoots per fragment at the start of the experiment was included in the design as a covariate to test for possible effects of initial size. The experimental unit term was pooled to increase the power of the analysis. Finally, permanova was performed to test for differences in morphological variables of transplanted fragments between sites (random factor) in the transplantation experiment, and a chi-square test was used to compare transplant survival between sites. Some fragments were lost in one site because of physical disturbance, thus there were 12 plants per site for morphological analyses.
Prior to permanova, data were normalized and dissimilarities calculated as Euclidian distances. Significance levels were calculated from 9999 permutations of the residuals under the reduced model. For some terms, there were not enough permutable units to get a reasonable test by permutation, so a P-value was obtained using a Monte Carlo random sample from the asymptotic permutation distribution (Anderson, Gorley & Clarke 2008). When a significant effect was found, post hoc pair-wise comparisons were used to distinguish between means, and separate univariate analyses (anovas) were performed on each response variable according to the permanova model. Prior to anova, normality and homogeneity were tested and data transformed where necessary to meet the assumptions of equal variances and normal distribution of residuals. When a significant effect was found, Fisher LSD test was used to test for differences among mean treatments. Multivariate and univariate analyses were performed using permanova+ for primer (Anderson, Gorley & Clarke 2008) and the r statistical program (version 2·12·2, R Development Core Team, 2011, free software).
In total, 36 seeds germinated, corresponding to 48% of the initial seed number. The majority of seeds (83·3%) initiated patch development after 2 years in culture and survived to the end of the experiment. Plants reached a maximum shoot production at 4–5 years (a mean of 164 shoots, corresponding to 2600 shoots per m2); thereafter, the production declined (Fig. 2). Shoot height increased with age similarly to shoot production, while the number of leaves per shoot reached a plateau at about 2 years (Fig. 2). At the end of the experiment, plants from rectangular containers had shoots higher than those from round containers (see Table S1 in Supporting Information). Plants extracted from rectangular containers in August 2009 (see Table S2), when they were 4 years old, had a large rhizome network (up to 7·8 m in length). Plants extracted 2 years later, in August 2011 were smaller (see Table S2), possibly reflecting a decline in plant vigour.
The first male flowers appeared in late April 2010, while the first female flowers were observed in mid-May 2010, when plants were 5 years old. Of the 16 plants remaining in the tanks, five females and two males flowered. The percentage of flowering plants in round (50%) and rectangular (37·5%) containers was similar (λ2 = 0·25, P =0·61). The mean densities of female and male flowers were 110 flowers m−2 (±45) and 33·3 flowers m−2 (±16·6 SE), respectively (see Table S3). Only one plant (male) that did not flower in the previous year flowered in the successive year (2011). The percentage of plants that flowered in the 2 years (2010 and 2011) in each of the two types of container was 50%. The mean sex ratio of flowering plants (0·37 ± 0·12) did not significantly deviate from an expected equal fraction (0·50) of male and female (P =0·72). No significant association was detected between flowering probability and plant size (λ2 = 0·49, P =0·48). Females were significantly larger than males in terms of number of shoots (F1,4 = 12·2, P =0·02; see Table S3). The flowering ratio of the shoots of female plants was also higher than that of males (F1,4 = 9·1, P =0·03; see Table S3), indicating a higher investment in reproductive structures in females. No developing fruit was found during the study period.
All fragments from mother plants produced in April and August 2009 generated new plants. In the spring experiment, plant morphology did not differ among experimental units (F3,12 = 2·6, P =0·52). After 5 months in culture (September 2009), the mean number of living shoots per plant was 25·6 (±2·2) and up to six new branches per plant (mean, 2·8 ± 0·4) were produced. The mean total rhizome length was 56·2 ± 3·7 cm, resulting in an approximately fourfold increase in the initial fragment length. Fragments produced in the fall experiment initiated vegetative expansion only in late spring of the subsequent year, June 2010. No significant effect of initial fragment size on plant morphology was observed (see Table S4). Plants regenerated from fragments without an apex had longer rhizomes than those from fragments with an apical meristem, but only for one genet, subsequently referred to as genet 1 (see Table S4; Fig. 3). For each fragment typology, a significant difference in rhizome length was detected between genets (Fig. 3, see Table S4): plants from fragments with an apex had higher shoots than those from fragments without an apex, but only for genet 2 (Fig. 3, see Table S4). For branch length, there was a significant effect of fragment type; fragments without an apex generated plants with branches longer than those with an apex (Fig. 3). No significant difference in the number of newly produced branches was observed among treatments (Fig. 3, see Table S4).
In the field at the time of planting, the number of shoots per cutting was similar across all sites (F1,30 = 0·332, P =0·569). Final transplant survival (Fig. 4) varied significantly between sites (100% vs. 75%, λ2 = 4·57, P =0·03). All cuttings that survived generated new plants. No differences in plant morphology between sites at the end of the observation period were detected (F1,22 = 5·59; P =0·32). Number of shoots increased during the summer of 2010 and declined in winter because of low temperatures, but in the following spring (May 2010) the number of shoots equalled those observed in the previous summer (Fig. 4). Up to three new branches were produced by plants, and the length of rhizome was approximately twice the initial cutting size (Fig. 5).
Previous studies have explored methods to maximize seed germination and seedling survival in order to produce alternative planting material to adult plants for seagrass restoration. The results of this study demonstrate that C. nodosa can be successfully established from seed and cultured up to 6 years in a nursery. Seed germination rate was high compared with field conditions (Orth et al. 2006a,b). Clonal patches increased in size up to 3–4 years of age, corresponding to the nonlinear patch growth behaviour described in demographic and modelling studies (Vidondo et al. 1997; Sintes et al. 2005), but remained smaller as compared with naturally established (Vidondo et al. 1997; Sintes et al. 2005), possibly because of the prolonged plant restriction caused by containers. The optimal container typology for seagrass culture has not been investigated to our knowledge. Here, both rectangular and round containers were suitable for growing C. nodosa, but the rectangular type provided plants with longer shoots. Our data further indicate that plants reached sexual maturity after a juvenile period of at least 5 years, but there was variability in the onset of reproduction among genets. Little is known about the factors that govern the onset of reproduction in seagrasses. There is evidence that transplanted seedlings of the fast-growing species Zostera marina (eelgrass) produce seeds in their first or second year (Orth & Moore 1986; van Katwijk et al. 2010). Because C. nodosa has a relatively extended life cycle (the oldest patch observed was 10·n5 years old, Vidondo et al. 1997), this could imply that during the initial years genets invested considerable energy in growth and colonization of space until they reach a size large enough to flower. Studies have shown that in some terrestrial plants, size rather than age triggers the onset of reproduction (Wesselingh et al. 1997). The lack of association between flowering and size, does not support the hypothesis of a threshold size for flowering in C. nodosa. Male flower density was within the range of that recorded in C. nodosa meadows (Terrados 1993) for the Mediterranean area, while that of female flowers was higher. Based on the number of ovaries per flower (two), the potential seed production for C. nodosa genets was ca. 220 seeds m−2. Unfortunately, no fruit was produced during the study period. Further studies would identify the conditions necessary for inducing seed maturation to maximize the number of seeds for future restoration efforts while minimizing seed collection pressure. This study also provides evidence for differences in growth capacity between sexes. Previous studies indicate that sex ratios can become female biased with increasing clone age in some species of the genus Phyllospadix, possibly due to sex differences in growth or survivorship (Williams 1995; Shelton 2008; Buckel et al. 2012). Here, the sex ratio of C. nodosa flowering genets was not biased towards females, and there was no difference in survival between sexes at least during the study period. However, females produced larger clones than males and invested more in sexual reproduction, producing three times as many reproductive structures as males. As females can grow faster, they could have a competitive advantage over males under limiting field growth conditions.
Results of the propagation experiments demonstrate that C. nodosa plants could be multiplied by cuttings from mother plants before decline in growth. At the maximum stage of vegetative development, a mother plant was sufficiently large to generate at least 50 new plants. These plants could be transplanted in the field or used to produce transplants for restoration. Fragments excised at the start of the growing season grew more rapidly than those obtained at the end of the growing season, providing plants as large as four times the initial cutting size in 5 months. Therefore, the season is an important factor to be considered in planning propagation programmes. Fragments without apical meristems generally produced plants with longer branches than those with apical meristems because of the lack of inhibitory effects of apical dominance on the vegetative development of lateral meristems (Terrados, Duarte & Kenworthy 1997). Therefore, it would be more convenient to use fragments without an apex as starting material to maximize the number of transplants available for restoration.
Most transplants obtained from stock plants were able to survive and grow in the natural habitat and produced patches within 4 months. Transplants, being vegetative progeny of adult clones, are better provisioned than seedlings and thus have a much higher probability of substrate colonization. Studies are needed to determine the number of mother plants required to maximize genetic diversity in restored populations. The success of restored populations is contingent upon the survival, growth and reproduction of the founding individuals, therefore ensuring the production of a sufficient number of female and male plants to transplant the correct sex ratio is extremely important in dioecious species. This issue has been largely ignored by managers until now.
This study is the first to provide guidance on how to create a nursery of plants from a limited number of seeds and is also the first reporting successful cultivation of a seagrass for the whole life history, from seed to mature individual. Previous cultivation efforts have obtained limited success, with a maximum permanence period of 2 years (Balestri, Vallerini & Lardicci 2010; Tanner & Parham 2010). Growing plants in a nursery for an extended period is more expensive than collecting plants and/or seeds from the field, requiring an investment in equipment, personnel and electricity. However, when choosing a restoration technique it is important to balance economic costs with ecological benefits. Traditional restoration techniques have not accounted for the impact of collection on existing populations. The main benefits of using nursery-grown plants as alternative or in addition to field-collected plants in seagrass restoration in terms of conservation of existing populations are evident when considering the number of seeds or plants required for achieving reasonable success in large-scale programmes. We estimated that the number of shoots and the rhizome network produced by stock plants from a mother plant (approximately 1300 shoots and rhizome network of 28 m) could be generated from two seeds. As each stock plant in turn supplied at least two transplants, approximately 100 transplants (in total 300 shoots) could be achieved from two seeds. Obtaining an equal shoot number through current seed-based restoration practices (Zarranz et al. 2010) would require at least 214 precultivated seedlings, corresponding to an initial number of 428 seeds, assuming a germination rate of 50% and no seedling mortality. Alternatively, approximately 30 m of rhizome would need to be removed from donor meadows if adult plants were used in restoration. Another important benefit of nursery-grown plants is the preservation of genetic diversity in restored populations, which provides biological insurance against fluctuations in environmental conditions and is a major goal of ecological restoration (Reynolds et al. 2012). Thus, the proposed system is a basic tool that can also be used for other seagrasses with low or unpredictable reproductive success, and the development of nurseries should be incorporated in future restoration programmes. This is currently the only sustainable methodology to produce plant material for transplanting programmes of species protected by legislation.
This study was partly funded by Solvay Chimica Italia S.p.A. of Rosignano Solvay (Italy). We thank Flavia Vallerini and Silvia Frosini for their technical assistance. We are grateful to Professor Philip Hulme, Professor Chris Frid, Doctor Chris Pickerel and an anonymous reviewer for valuable suggestions and comments that helped improve and clarify earlier versions of the manuscript.