Density-Dependent Regulation in an Invasive Seaweed: Responses At

JEC_720.fm Page 820 Monday, September 23, 2002 11:06 AM

Journal of Blackwell Science, Ltd Ecology 2002 Density-dependent regulation in an invasive seaweed: 90, 820–829 responses at plant and modular levels

FRANCISCO ARENAS†, ROSA M. VIEJO* and CONSOLACIÓN FERNÁNDEZ Unidad de Ecología, Departamento de Biología de Organismos y Sistemas, Universidad de Oviedo, Asturias, Spain

Summary 1 The effects of density on the vital rates, growth form and population size structure of the invasive seaweed Sargassum muticum (Yendo) Fensholt were evaluated experimentally. 2 The initial difference in plant number between the highest and the lowest density (200 and 6400 plants m−2) was considerably reduced by the end of the experiment. Surpris- ingly, this was mostly due to numbers increasing at lower densities, probably because microscopic forms were not removed during the experimental thinning. 3 The allometric length–dry mass relationship fitted a simple linear model on log-log scale for both the highest and the lowest densities, but had different slopes. At higher densities plants became taller and thinner as a consequence of variations in the production and growth of modules. 4 Mean size (dry mass) and the development of size hierarchies of plants were affected by both the addition of further (microscopic) recruits and asymmetrical competition among plants. Plant length distributions were also influenced by changes in the growth form of plants. The length hierarchies of main branches also suggested the presence of asymmetrical competition at this modular level. 5 Density influenced both mean size and morphology of the plants and thus induced changes in reproduction. The negative effect of density on individual plant size reduced the percentage of fertile plants and possibly their annual reproductive allocation, but these effects may be attenuated by morphological responses. 6 The responses of S. muticum to crowding are closely linked to its ability to colonize bare space. The massive reproductive output and very limited dispersal range account for local and dense recruitment patterns. Our results suggest that the responses of S. muticum to crowding allow the establishment of dense populations with high persistence and resistance to colonization by other species. Key-words: allometry, growth form, intraspecific competition, Sargassum muticum, size hierarchies Journal of Ecology (2002) 90, 820–829

reproductive inequalities among neighbours (e.g. Introduction Weiner 1985; Weiner & Thomas 1986; Schmitt et al. The impact of density on the population dynamics of 1987; Rice 1990; Weiner 1990; but see Turner & Rab- terrestrial plants has been extensively investigated over inowitz 1983; Ellison 1987). the past few decades (e.g. Harper 1977; Antonovics & The modular construction of plants enables indi- Levin 1980; Thomas & Weiner 1989; Rice 1990; Weiner viduals to respond to crowding not only by changes in 1990; Yastrebov 1996). Some of the best-characterized growth rates, but also by plastic changes in their archi- population responses to increased plant density are the tectures. Thus, many species display stem elongation reduction of vital rates and the development of size and and reduced branching when they grow at higher densities (see Hutchings & de Kroon 1994). Within popu- *Correspondence and present address: Dipartimento di Sci- lations, these changes in the shape of plants are enze dell’Uomo e dell’Ambiente, Via A. Volta 6, I-56126 Pisa, commonly dependent on their ranking in a size hier- Italy (fax +39 05049694; tel. +39 050500943/500018; e-mail [email protected]). archy, with a higher relative elongation of small © 2002 British †Present address: Marine Biological Association of the UK, ‘suppressed’ plants in relation to large ‘dominant’ ones Ecological Society The Laboratory, Citadel Hill, Plymouth, UK. (Geber 1989; Weiner et al. 1990; Weiner & Thomas

JEC_720.fm Page 821 Monday, September 23, 2002 11:06 AM

821 1992; Yokozawa & Hara 1995; Berntson & Wayne Density- 2000). The morphological changes have often been dependence in an shown to be adaptive in terms of growth and reproduc- invasive seaweed tion (Schmitt et al. 1995; Dudley & Schmitt 1996; Dorn et al. 2000). Density dependence has been considerably less investigated in seaweed stands, although the results of manipulative experiments over the last decade indicated that algal responses to crowding are often similar to those observed in terrestrial plants. Hence, negative effects of density on survival and growth are common in algal populations (Reed 1990; Ang & De Wreede 1992; Kendrik 1994; Creed et al. 1996, 1998). Never- theless, the effect of density on the reproductive output of macroalgae is still poorly documented, as is the development of size and reproductive hierarchies (but see Schiel 1985; Reed 1990; Creed et al. 1998). Dynam- Fig. 1 Scheme of the typical morphology of a Sargassum ics of modules have been investigated in a few seaweeds, muticum plant. Dmb = dominant (longest) main branch; mostly clonal red algae (e.g. Santos 1995; Scrosati & Mb = main branch; Sb = secondary branch; S = stem; H = holdfast; R = reproductive structure (androgynous receptacle); Servière-Zaragoza 2000) but studies relating genet Av = air vesicle; L = leave-like structures. competition to clone dynamics are rare (see Lazo & Chapman 1998; Viejo & Åberg 2001). Moreover, to our knowledge the interplay of density, growth form and 1996. The invasive brown seaweed S. muticum was first size inequality has not yet been explored in seaweeds. detected in this area in 1988 and its percentage cover in In this study we investigated density dependence in a the low intertidal has been close to 100% since 1991 brown seaweed with high invasive potential, Sargassum (Fernández 1999). muticum (Yendo) Fensholt. This is a Japanese species On European coasts, S. muticum exhibits a very that rapidly spread after attached individuals were first characteristic annual growth-cycle, although there are found in Europe in the 1970s (Critchley et al. 1983). Its some differences in phenology among areas (see localized pattern of dispersal (Deysher & Norton 1982) Jephson & Gray 1977; Givernaud et al. 1991; Fernández helps the quick establishment of S. muticum popula- 1999). In the north of Spain, main branches arise from tions once individuals have colonized a new habitat. the stem during autumn and grow throughout winter The species occupies mainly sheltered shores, where it and spring until the onset of the reproductive period, in often forms dense monospecific stands (Jephson & late spring. The reproductive period ends in August Gray 1977; Critchley et al. 1983; Fernández et al. 1990). (Fernández 1999) and only the holdfast and stem The growth form of S. muticum is modular and remain as perennial structures. approaches the structural complexity of terrestrial In June 1995, 50 concrete floor tiles (30 × 30 × 5 cm) plants (Chamberlain et al. 1979; Yoshida 1983). A were seeded with S. muticum propagules in situ, by plant (genet) of S. muticum is attached to the substra- immersing them for 2 months under a canopy of repro- tum by a perennial holdfast that gives rise to a single ductive plants in a flat area of the bay, 0.2–0.3 m above stem (Fig. 1). Every year, several apically extending the lowest tidal level. Plants used as inoculum were main branches emerge from the stem and produce well- subsequently removed from the area. Tiles that were developed secondary branches (Fig. 1), which in turn profusely colonized by propagules of S. muticum may give rise to higher-order branches. (n = 40) were used in the experiment. In November We examined the effects of density on the vital rates 1995, when recruits were 1–2 cm long, tiles were ran- of S. muticum, as well as the development of size (dry domly assigned to five different experimental density mass) and length hierarchies. The morphological treatments (eight tiles per density) equivalent to 200, responses to crowding (i.e. production and growth of 400, 800, 3200 and 6400 recruits m−2. In the studied modules) and its relationship with size distributions area, average natural densities of recruits (estimated were also analysed. Modularity was considered at the from sampled plots of 0.25 m2) ranged from 761 to 974 scales of main (primary) and secondary branching. plants m−2 (Fernández 1999; Arenas & Fernández 2000), with a maximum value of 12 380 plants m−2 (F. Arenas, unpublished data). Recruits were thinned to Materials and methods the required initial density by dividing the surface of ×  - the tile into 18, 5 10 cm rectangles and scraping off recruits with forceps to leave only 1, 2, 4, 16 and 32 © 2002 British Ecological Society, The field experiment was carried out in Aramar, a shel- individuals per rectangle. Journal of Ecology, tered bay near Cape Peñas, on the north coast of Spain Every effort was made to remove individuals com- 90, 820–829 (43°36′41″N, 5°46′29″W) from June 1995 to June pletely, by gouging and scraping the tiles. Tiles were

JEC_720.fm Page 822 Monday, September 23, 2002 11:06 AM

822 placed 200 cm apart to avoid physical interaction bears hundreds of thousands of receptacles (Norton F. Arenas, between plants from different tiles. These plots simu- 1977; Norton & Deysher 1989), often a few millimetres R. M. Viejo & lated the clumped distribution of S. muticum at Ara- in size. Reproductive allocation was therefore assessed C. Fernandez mar, where plants are frequently found attached to only for secondary branches and only at one low (400 boulders immersed in sand or embedded in mud. plants m−2) and one high (6400 plants m−2) density. Sec- Zones adjacent to the plots were kept clean of adult ondary branches (30–40, from a total of seven to eight plants throughout the course of the experiment. plants at each density) were randomly selected from the In order to study the effect of density on survival, the longest main branch of plants, their receptacles separ- number of plants per plot was assessed bimonthly from ated and the total dry mass of both receptacles and November 1995 to June 1996. All plants were counted vegetative tissue measured. in the plots at the lowest experimental densities (200 −2 −2 and 400 plants m ) but at 800, 3200 and 6400 plants m   estimates were derived from counts in 10 subsamples (5 × 5 cm), except in May and June 1996, when all indi- The change in plant number in the plots between two viduals were counted. Working at low intertidal levels consecutive times was expressed as a percentage. Vari- imposes logistic constraints and for each census there- ation in plant number among treatments (i.e. experi- fore we sampled only four to seven randomly selected mental densities) was analysed using an analysis of tiles per density. variance with density as a fixed factor. Cochran’s tests Light has been cited as a major factor limiting the were used to test data for homogeneity of variances growth and survival of competing seaweeds (e.g. Ang & (Winer et al. 1991). Differences among means after De Wreede 1992; Creed et al. 1998). Total daily irradi- significant F-tests were analysed by Student-Newman- ance data (MJ m−2) for the period of November 1995 to Keuls (SNK) tests. June 1996 were kindly provided by the nearest meteor- To investigate the effect of density on growth form, ological observatory (CMT Asturias, Oviedo, Spain), the allometric relationship between individual length located at about 28 km from the study site. and dry mass was analysed for the lowest and the The impact of density on the mean dry mass and highest experimental densities (i.e. 200 and 6400 length of plants, and on size distributions, was invest- plants m−2) in the plots collected in May. A random igated in May 1996, at the end of the growing season, sample of 270 plants of at least 1 cm length was taken and June 1996, when plants became reproductive. In per density. For each density, the log (dry mass)–log May, half of the plots (four for each density) were (length) data were fitted to both linear and quadratic collected from the shore and transported to the regressions (standard least square regressions with laboratory. Plants were then carefully removed, rinsed plant length as the independent variable). The slopes in fresh water, and total length of each individual was of the linear models were compared by analysis of measured to the nearest millimetre, and number and covariance. Analyses of covariance were also used length of main branches noted, before drying for 48 h for testing the effects of density on: (i) plant length at 60 °C and weighing (± 0.0001 g). (plant dry mass as covariate); (ii) dry mass of the long- Plastic morphological responses to density were est main branch (plant dry mass as covariate); and assessed in some larger plants in which they were more (iii) number of secondary branches of the longest main evident. Five well-developed individuals (> 0.07 g dry branch (covariate: length of the main branch) in mass) were thus selected from each of the May-collected the 20 sampled individuals (see above). Slopes were plots (20 per density) and (i) the length and dry mass of homogeneous in all the cases. Student’s t-tests and the longest main branch, (ii) number, length and dry Bonferroni corrections were used for posthoc comparison mass of secondary branches on this main branch, and on adjusted means after significant F-tests. Finally, an (iii) length and dry mass of the rest of main branches, analysis of variance was used for testing the effect of were measured for each plant. density on number of main branches per plant (there The rest of the plots were collected in June, coincid- was no relationship between this variable and plant ing with the onset of reproduction. One plot was lost size) and differences among means were analysed by from the three lowest densities (200, 400 and 800 plants a SNK test. m−2). Each plant in the remaining three to four plots per In order to examine the hierarchy in the dry mass density was counted and assessed for reproductive and length distributions of both plants and main status, total length and number and length of main branches, unbiased Gini coefficients (Weiner & Solbrig branches. To investigate if crowding has an impact on 1984) were calculated for each plot. This coefficient is a the probability and/or timing of reproduction, the statistical measure of inequality in the size distribution, numbers of both fertile plants and plants having imma- having a minimum value of 0, where all the individuals ture reproductive structures (receptacles) were counted, (or branches) are equal, and a theoretical maximum of in order to estimate the proportion of plants that would 1, where all individuals but one have size 0. Analyses of © 2002 British Ecological Society, become fertile later in the reproductive season. variance were used to compare the Gini coefficients Journal of Ecology, The separation, counting and weighing of recept- at different densities and times (May, June). When 90, 820–829 acles is very time-consuming because a single plant possible, posthoc pooling procedures were used to

JEC_720.fm Page 823 Monday, September 23, 2002 11:06 AM

823 remove the non-significant P > 0.25 interaction term Table 1 Analysis of variance for the effects of initial densities Density- density × time, to increase the power of the test for the on plant number during May to June 1996. As counts were dependence in an main factors (Underwood 1997). done in different plots each time, the measures were independent and the factor time was included in the analysis. invasive seaweed Analysis of covariance was used to test the effect of The dependent variable was the log proportion of final to crowding in the reproductive allocation of secondary initial densities (initial in November, final in May or June, branches. Furthermore, the relationship between respectively). The interaction density × time was not signi- reproductive and vegetative dry mass was examined to ficant (P = 0.397) and pooled residual MS was calculated. test if reproductive allocation of secondary branches Variances were homogeneous increased with branch size. The value of the y-intercept Source of variation d.f. MS FP and its 95% confidence limits were calculated using standard least-square linear regression to evaluate if Density 4 0.387 8.06 0.000 the y-intercept was significantly < 0 (see Samson & Time 1 0.204 4.25 0.048 Werk 1986; Klinkhamer et al. 1990). Residual 31 0.048 Data from three missing plots were replaced by the mean Results value of the group, and 3 d.f. subtracted from residual.

     plots thinned to 400 plants m−2 increased in density by The number of plants in the plots exhibited successive an average of 5.6%. phases of reduction and increase over time (Fig. 2a). Surprisingly, the number of plants visible increased The number of plants declined in most of the treatments from January to May in all treatments, as did irradiance from November to January (Fig. 2a), when irradiance levels (Fig. 2). These increases were density-dependent

levels were minimal (Fig. 2b). Survival over this period from January to March (analysis of variance, F4,21 = declined with initial experimental density (signiﬁcant 16.38, P < 0.001), being greatest at the lowest initial

effect of density, analysis of variance, F4,21 = 9.82, density (Fig. 2a; SNK test, 200 > 800 > 400 c. 3200 c. P < 0.001), from 91.7% at the lowest density treatment 6400 plants m−2). (200 plants m−2) to 60.3 at the highest one, although In the last period, May to June, all the treatments again experienced reductions in plant number (Fig. 2a), although this mortality was not density-dependent (Table 1, non-signiﬁcant interaction, density × time). Over the whole period, net mortality was only observed at the two highest initial densities, the remainder show- ing a net increase (Fig. 2a). The initial 32-fold difference among the highest and the lowest densities (6400 vs. 200 plants m−2) was reduced to eight by the end of the experiment (4444 vs. 551 plants m−2, mean values).

   

Density-dependent changes in the architecture of plants were evident in May. The allometric relationship between length and dry mass changed with density (Fig. 3). This relationship fitted a linear regression in a log-log plot for both the high and the low densities tested (r2 = 0.904 and 0.903, respectively, P < 0.0001 in both cases). Although quadratic regressions were also highly significant (P < 0.0001), they only slightly increased the amount of variance explained in relation to the linear model (< 1.5% for either density). The slope of the linear model (log dry mass = a + b log length) was lower for the high density (1.28 vs. 1.43; × analysis of covariance, density length, F1,535 = 15.64, P < 0.001). This indicated that plants were more elongated at the highest density and differences between Fig. 2 (a) Temporal variation in the number of plants for the densities became more evident in large plants (Fig. 3). ± different experimental densities. Mean SE values are shown, In fact, when large plants (> 0.07 g dry mass) of all © 2002 British number of plots equals four in January, May and June and Ecological Society, seven in November and March, except where indicated. (b) five experimental densities were compared, we Journal of Ecology, Temporal variation in total daily irradiance from November observed a gradual elongation (Fig. 4a, Table 2), a 90, 820–829 1995 to June 1996 (monthly mean ± SE). decrease in the number of main branches (Fig. 4b;

JEC_720.fm Page 824 Monday, September 23, 2002 11:06 AM

824 analysis of variance, F4,15 = 9.13, P < 0.001) and a F. Arenas, higher dominance of the longest main branch (i.e. R. M. Viejo & relatively higher dry mass of this branch) (Fig. 4c, C. Fernandez Table 2) from the lowest to the highest density. At high density, the longer dominant main branch held a higher number of secondary branches (Fig. 4d), but these were sparsely distributed so that there were fewer of them if main branches of similar lengths were compared (Fig. 4d, Table 2). Overall, the allometric length–dry mass relationship of S. muticum changed as a consequence of alterations in the production and growth of modules. Hence, the number of main branches produced per plant decreased, whereas size inequality among main branches increased Fig. 3 Length–dry mass relationship for plants growing at the with density. The frequency of secondary branching in −2 highest (6400 individuals m , ) and the lowest (200 the longest main branch was reduced but the lengths of −2 individuals m , ) experimental densities in May. these secondary branches increased with density, as plants became elongated. There was a signiﬁcant correlation between length of the dominant main branch and mean length of secondary branches (Pearson’s correlation coefﬁcient, r = 0.853, P < 0.001, n = 100). The allometric length–dry mass relationship of the secondary branches, however, was not density-dependent (data not shown).

         

The mean dry mass of the plants in May increased as the density decreased, except for the lowest experimental density (200 plants m−2), where mean dry mass

dropped again (Fig. 5; analysis of variance, F4,15 = 4.20, P = 0.018). The values of the Gini coefficients for the size distributions (an indication of size inequality) did not differ significantly among densities (analysis of variance, F = 1.05, P = 0.415), although they tended Fig. 4 Morphological characteristics of plants growing at the 4,15 five experimental densities (1 = 200, 2 = 400, 3 = 800, 4 = 3200, to increase with density if the lowest density was 5 = 6400 plants m−2) as mean values (± SE), n = 20. (a) length excluded (Fig. 5). The patterns of mean weights and of the plants; (b) number of main branches; (c) dry mass of the inequalities were related to the percentage of small longest main branch; (d) number of secondary branches (in plants in the plots. Small plants were the most frequent the longest main branch). Dots in (a), (c) and (d) indicate adjusted in all plots, but the highest percentages were observed mean values after analysis of covariance (see Table 2). In (a) and ± (c) the adjusted values were back-transformed from log data at the highest density, where 79.3% 1.7 had dry and the associated statistical bias corrected (see Newman 1993 mass < 0.05 g (mean values ± SE, n = 4). The value for method). Letters indicate the groups differentiated by fell with density to 52.4% ± 2.0 at 400 plants m−2, before posthoc tests where these were significant at P < 0.05. increasing to 73.1% ± 9.4 at the lowest density. Maximum

Table 2 Analyses of covariance for the effect of initial densities on different morphological attributes of larger plants (> 0.07 g). Plant length and dry mass of plants and branches were log–transformed. The interaction between the covariate and the treatment (density) was not signiﬁcant in any case

Source of Dependent variable variation d.f. MS FP

Plant length Density 4 0.115 24.92 0.000 (covariate: dry mass of the plant) Residual 94 0.005 Longest main branch Dry mass of the branch Density 4 0.104 17.24 0.000 © 2002 British (covariate: dry mass of the plant) Residual 94 0.006 Ecological Society, Number of secondary branches Density 4 412.084 4.04 0.005 Journal of Ecology, (covariate: length of the branch) Residual 94 102.134 90, 820–829

JEC_720.fm Page 825 Monday, September 23, 2002 11:06 AM

825 Density- dependence in an invasive seaweed

Fig. 5 Mean dry mass per plant in May for the ﬁve experimental densities (codes as in Fig. 4) Gini coefﬁcients (dots) for plant size distributions. Mean values ± SE are shown, n = 4. Letters indicate the groups with different dry mass at P < 0.05.

dry mass, however, did not differ among densities (F4,15 = 0.54, P = 0.710). The mean length of plants in the plots collected in May and June followed a similar trend to the mean weights, although the differences were no longer signi- Fig. 6 Mean lengths for plants (a) and main branches (b) at ficant (Fig. 6a, Table 3). Gini coefficients for plant the five experimental densities and their Gini coefficients length distributions increased steadily with density and (dots) (June data, patterns were similar in May; Tables 3 and were significantly higher at the highest density (Fig. 6a, 4, non-significant interaction density × time). Mean values ± Table 3) (Pearson’s correlation coefficient among log SE are shown, n = 3 for the three lowest densities, n = 4 for the density and Gini coefficient, r = 0.757, P < 0.001, rest. Letters indicate different groups for Gini coefficients at P < 0.05 (pooled May and June data). n = 17 in June). This pattern was due to the fact that density-dependent elongation was more evident in large plants (Fig. 3). The contrasting results shown in immature receptacles were included to give an estimate Fig. 6(a), where mean values for all plants are shown, of the percentage in late summer, the trends were and Fig. 4(a), which considers only large plants, can be similar (Fig. 7a). similarly explained by the dominance of small plants at The reproductive allocation of the secondary the lowest and the highest densities. branches was apparently not affected directly by den- The effect of density on the size structure of the sity (Fig. 7b; analysis of covariance for reproductive

stands was also investigated at the modular level, by dry mass, F1,67 = 3.28, P = 0.075). The relationship analysing the length distributions of all the main between vegetative and reproductive dry mass fitted a branches of all plants. Again, patterns were not signi- linear regression (r2 = 0.853, P < 0.001) with a negative ficantly different among densities but the Gini coeffi- y-intercept (y-intercept ± 95% confidence limits = cients increased with density (Fig. 6b, Table 4). −0.0166 ± 0.003; see Fig. 7b), indicating that reproductive allocation was an increasing function of the size of      secondary branches. In order to assess the reproductive allocation at the The percentage of fertile plants in June was not signi- level of plant, it is important to note that although the ficantly different among densities (analysis of variance, total number of secondary branches per individual did

F4,12 = 1.38, P = 0.299) although, like mean weights not differ among densities (analysis of covariance,

and lengths, they tended to be higher at lower densities, F1,12 = 0.007, P = 0.936, plant dry mass as covariate), except at the lowest one (200 plants m−2), where the there were plastic changes in growth form (number and percentages dropped (Fig. 7a). When plants with inequality of main branches) and plants of similar sizes

Table 3 Analysis of variance for the effects of experimental densities on mean plant length and Gini coefﬁcients for plants collected in May and June 1996. The interaction density × time was not signiﬁcant (P = 0.973 in both cases) and pooled residual MS were calculated. Variances were homogeneous

Mean plant length Gini coefﬁcient

Source of variation d.f. MS FPMS FP

Density 4 71.179 1.55 0.212 0.024 5.30 0.002 © 2002 British Time 1 4859.710 100.12 0.000 0.017 3.72 0.063 Ecological Society, Residual 31 45.843 0.005 Journal of Ecology, 90, 820–829 Data from three missing plots were replaced by the mean value of the group, and 3 d.f. subtracted from residual. JEC_720.fm Page 826 Monday, September 23, 2002 11:06 AM

826 Table 4 Analysis of variance for the effects of experimental densities on mean length of main branches and Gini coefficients of F. Arenas, the length branch structure. The plots were collected at two different times: May and June 1996. The interaction density × time R. M. Viejo & was not significant and pooled residual MS were calculated. Variances were homogeneous C. Fernandez Mean length Gini coefficients

Source of variation d.f. MS FPMS FP

Density 4 50.229 1.67 0.517 0.009 4.38 0.006 Time 1 2107.171 70.16 0.001 0.089 41.27 0.000 Residual 31 30.033 0.002

Data from three missing plots were replaced by the mean value of the group, and 3 d.f. subtracted from residual.

& Santelices 1991 for a review), and S. muticum also has an extraordinary capacity for regenerating from small fragments of holdfasts (Fletcher & Fletcher 1975). These sources of recruits could play a similar role to that of seed or seedling banks in terrestrial plant populations, reducing the amplitude of the fluctuations about an equilibrium population (MacDonald & Watkinson 1981), as described by Ang (1991) and Creed et al. (1996) for two seaweed species. The existence of such reserves may enable successful invaders like S. muticum Fig. 7 (a) Percentages of fertile and potentially fertile to recover from canopy loss following disturbance. (hatched and open bars, respectively) plants in June for the five experimental densities. Mean values ± SE are shown, n = 3 for the three lowest densities, n = 4 for the two highest        densities. (b) Relationship between the vegetative and reproductive dry mass of secondary branches at 400 () and Density-dependent changes in the average form of 6400 () plants m−2. plants have been detected in some kelp species, which exhibit stipe elongation in crowded stands (Hymanson et al. 1990; Reed 1990; Holbrook et al. 1991; Sjøtun had longer secondary branches in the dominant main et al. 1998). To our knowledge, however, this is the first branch when growing at higher densities. marine study where the effect of competition on plant form was also investigated allometrically and where plastic changes in morphology were related to the Discussion modular character of the alga. Responses of the invasive seaweed S. muticum to Density changed the slope of the allometric length– crowding appear remarkably similar to those of higher dry mass relationship in S. muticum, with plants grow- plants. Plastic changes in the birth rate and growth of ing at higher densities becoming taller and thinner modules that regulate the accumulation of biomass, (Fig. 3). In terrestrial systems the log-log allometry and reduced genet mortality, may allow the establish- often becomes non-linear in crowded stands, as a result ment and persistence of highly dense populations, of a higher relative elongation of small ‘suppressed’ despite the presence of asymmetric competition. plants in relation to large ‘dominant’ ones (e.g. Geber 1989; Weiner & Thomas 1992; Yokozawa & Hara 1995;     : Berntson & Wayne 2000). This was not observed in S. muticum, where plants changed shape across densities,     ‘   but the allometric growth pattern was similar for plants ’ of different size within a stand. Net mortality over the whole experimental period was The birth rate of modules (main branches) decreased only detected at the two highest densities. The number with density (Fig. 4). Such regulation may prevent of visible plants increased strikingly, especially in the overcrowding in clonal phanerogams with tightly lowest density between January and May, when irradi- aggregated modules (i.e. with a ‘phalanx’ growth form; ance levels rose (Fig. 2). In this region, S. muticum Lovett Doust 1981; Schmid & Harper 1985; de Kroon is only fertile in late spring and summer (Arenas & & Kwant 1991) and in some bryophytes (van der Fernández 1998), and this increase must therefore be due Hoeven & During 1997). The results of this and a pre- to growth of germlings that were not visible during the vious study on the brown alga Ascophyllum nodosum experimental thinning, or non-removal of microscopic (Viejo & Åberg 2001), demonstrated that similar © 2002 British Ecological Society, plant fragments. The presence of a bank of microscopic mechanisms could take place in seaweeds and could Journal of Ecology, germlings with a delayed development has been explain the absence of self-thinning in several species of 90, 820–829 demonstrated for many seaweed species (see Hoffman clonal red algae (Scrosati & Servière-Zaragoza 2000). JEC_720.fm Page 827 Monday, September 23, 2002 11:06 AM

827 Elongation of the dominant main branch and a (e.g. Samson & Werk 1986). The mean size of S. muticum Density- reduction of the frequency of secondary branching plants decreased with density, with the exception of the dependence in an at high density (Fig. 4) may be similar to the ‘shade lowest density, and this was related to the percentage of invasive seaweed avoidance responses’ (Smith 1982) of terrestrial plants. fertile plants (Fig. 7a). Furthermore, the reproductive Such responses to crowding are adaptive (Schmitt allocation of secondary branches was size-dependent, et al. 1995; Dudley & Schmitt 1996; Dorn et al. 2000) which suggested the same pattern for the whole plants and are mediated by phytochrome (Smith 1982; Casal within a stand (see also Arenas & Fernández 1998). & Smith 1989). Algal canopies are similarly exposed to However, comparison among densities is confounded changes in light quality, which could be detected by by differences in both size and morphology. Plants of their phytochrome-like photoreceptors (Salles et al. identical dry mass growing at higher densities exhibited 1996). a longer main branch with longer secondary branches, which would tend to diminish the size-dependent      differences in reproduction.  -   The trend for dry mass inequality to increase with      density supports the presence of asymmetrical com- S. MUTICUM petition (see also Weiner 1985; Weiner & Thomas 1986 for terrestrial plants; Reed 1990; Creed et al. 1998 for The responses of S. muticum to crowding are closely seaweeds). Light competition appears to be critical in linked to the way this aggressive invader (Norton 1976) establishing the pattern of dominant and suppressed colonizes and occupies new habitat. It is self-fertile and individuals in both terrestrial and marine systems has effective means of long-range dispersal, as drifting (Schmitt et al. 1987; Creed et al. 1998). Nutrients, fragments become fertile while suspended. The which are above- rather than below-ground resources embryos of S. muticum, however, are fast-sinking, for seaweeds, may, however, also be more readily multicellular propagules, which account for especially available for larger plants, as reduced water movement dense local colonization patterns (Deysher & Norton around small fronds can lead to them being nutrient 1982). The growth form of S. muticum is similar to the limited (Denny 1988). phalanx strategy of terrestrial clonal plants, with a very The pattern is actually reversed at the lowest density compact structure of closely spaced ramets (Lovett (Fig. 5), but this was probably the consequence of Doust 1981). It combines the fast growth of the ramets recruit addition, rather than a reﬂection of reduced (main branches) with the persistence of the perennial performance. In seaweed stands, negative effects at basal part (Norton 1976). The morphological plasti- lower densities have occasionally been reported (e.g. city of the species in the production of modules (main Hay 1981; Schiel 1985; Bertness & Leonard 1997) and branches) and branch elongation reduces interference related, for species living at low intertidal and subtidal among neighbours, allowing plants to grow in locally levels, to increases in physical battering and abrasion dense populations with low mortality and diminishing (Schiel 1985). Our experiment was, however, conducted the effects of density on reproduction. S. muticum also in a wave-protected area and the percentages of broken exhibits mechanisms to compensate for eventual plants were similar at 200 and 400 plants m−2 (5.1% and canopy losses, further enabling rapid colonization, and 6.2%, respectively). consolidation and persistence of local populations, The progressive elongation of S. muticum as density preventing invasion by other species. increases affected all plants within a stand and there is therefore a spreading of the length distributions and a Acknowledgements progressive increase in length inequality as density increases (Fig. 6a). The distributions of main branch We thank David Gutiérrez, Victoria López-Dóriga, lengths suggested that asymmetric competition was Emilio Marañón, Marcos Méndez and Rosa Menéndez also operating at the modular level (Fig. 6b), which for their help with experimental manipulation in the is related to the functional independence of main ﬁeld. We are also grateful to Lindsay Haddon, Hans de branches within a plant (Chamberlain et al. 1979). Such Kroon, Brezo Martínez, Ricardo Scrosati, Jacob length effects may be harder to detect in terrestrial Weiner and two anonymous referees for their valuable plants (e.g. Geber 1989 and references therein), because critical comments on the manuscript. This study was elongation commonly affects small individuals more funded by the EU MASTIII project EUROROCK than large individuals. (MAS3-CT95-0012).

     References © 2002 British Ecological Society, In terrestrial systems the effects of density on the Ang, P.O. (1991) Natural dynamics of a Fucus distichus Journal of Ecology, reproductive output and reproductive effort are often (Phaeophyceae, Fucales) population: reproduction and 90, 820–829 indirect, mediated by changes in the size of the plants recruitment. Marine Ecology Progress Series, 78, 71–85. JEC_720.fm Page 828 Monday, September 23, 2002 11:06 AM

828 Ang, P.O. & De Wreede, R.E. (1992) Density-dependence in a Hay, M. (1981) The functional morphology of turf-forming F. Arenas, population of Fucus distichus. Marine Ecology Progress seaweeds: persistence in stressful marine habitats. Ecology, R. M. Viejo & Series, 90, 169–181. 62, 739–750. Antonovics, J. & Levin, D.A. (1980) The ecological and van der Hoeven, E.C. & During, H.J. (1997) The effect of C. Fernandez genetic consequences of density-dependent regulation in density on size frequency distributions in chalk grassland plants. Annual Review of Ecology and Systematics, 11, 411– bryophyte populations. Oikos, 80, 533–539. 452. Hoffman, A.J. & Santelices, B. (1991) Banks of algal micro- Arenas, F. & Fernández, C. (1998) Ecology of Sargassum scopic form: hypothesis on their functioning and com- muticum (Phaeophyta) on the North Coast of Spain. III. parison with seed banks. Marine Ecology Progress Series, Reproductive ecology. Botanica Marina, 41, 209–216. 79, 185–194. Arenas, F. & Fernández, C. (2000) Size structure and dynam- Holbrook, N.M., Denny, M.W. & Koehl, M.A.R. (1991) ics in a population of Sargassum muticum (Phaeophyceae). Intertidal ‘trees’: consequences of aggregation on the Journal of Phycology, 36, 1012–1020. mechanical and photosynthetic properties of sea-palms Berntson, G.M. & Wayne, P.M. (2000) Characterizing the size Postelsia palmaeformis Ruprecht. Journal of Experimental dependence of resource acquisition within crowded plant Marine Biology and Ecology, 146, 39–47. populations. Ecology, 81, 1072–1085. Hutchings, M.J. & de Kroon, H. (1994) Foraging in plants: Bertness, M.D. & Leonard, G.H. (1997) The role of positive the role of morphological plasticity in resource acquisition. interactions in communities: lessons from intertidal habi- Advances in Ecological Research, 25, 59–238. tats. Ecology, 78, 1976–1989. Hymanson, Z.P., Reed, D.C., Foster, M.S. & Carter, J.W. Casal, J.J. & Smith, H. (1989) The function, action and adap- (1990) The validity of using morphological characteristics tive signiﬁcance of phytochrome in light-grown plants. as predictor of age in the kelp Pterygophora californica Plant, Cell and Environment, 12, 855–862. (Laminariales, Phaeophyta). Marine Ecology Progress Chamberlain, A.H.L., Gorham, J., Kaen, D.F. & Lewey, S.A. Series, 59, 295–304. (1979) Laboratory growth studies on Sargassum muticum Jephson, N.A. & Gray, P.W.G. (1977) Aspects of the ecology (Yendo) Fensholt. Botanica Marina, 22, 11–19. of. Sargassum Muticum (Yendo) Fensholt, in the Solent Creed, J.C., Kain, J.M. & Norton, T.A. (1998) An experimen- region of the British Isles. I. The growth cycle and epiphytes. tal evaluation of density and plant size in two large brown Biology of Benthic Organisms. 11th European Symposium seaweeds. Journal of Phycology, 34, 39–52. of Marine Biology (eds B.F. Keegan, P.O. O’Ceidigh & Creed, J.C., Norton, T.A. & Kain, J.M. (1996) Are neighbours P.J.S. Boaden), pp. 367–375. Pergamon, Oxford. harmful or helpful in Fucus vesiculosus populations? Kendrik, G.A. (1994) Effects of propagule settlement density Marine Ecology Progress Series, 133, 191–201. and adult canopy on survival of recruits of Sargassum spp. Critchley, A.T., Farham, W.F. & Morrell, S.L. (1983) A chro- (Sargassaceae: Phaeophyta). Marine Ecology Progress nology of new European sites of attachment for the invasive Series, 103, 129–140. brown alga, Sargassum muticum 1973–81. Journal of the Klinkhamer, P.G., Jong, T.J. & Meelis, E. (1990) How to test Marine Biological Association of the United Kingdom, 63, for proportionality in the reproductive effort of plants. 799–811. American Naturalist, 135, 291–300. Denny, M.W. (1988) Biology and the Mechanics of the Wave- de Kroon, H. & Kwant, R. (1991) Density-dependent growth Swept Environment. Princeton University Press, Princeton. responses in two clonal herbs: regulation of shoot density. Deysher, L. & Norton, T.A. (1982) Dispersal and coloniza- Oecologia, 86, 298–304. tion in Sargassum muticum (Yendo) Fensholt. Journal of Lazo, M.L. & Chapman, A.R.O. (1998) Components of Experimental Marine Biology and Ecology, 56, 179–195. crowding in a modular seaweed: sorting through the con- Dorn, L.A., Hammond-Pyle, E. & Schmitt, J. (2000) Plasticity tradictions. Marine Ecology Progress Series, 174, 257–267. to light cues and resources in Arabidopsis thaliana: testing Lovett Doust, L. (1981) Population dynamics and local for adaptive value and costs. Evolution, 54, 1982–1994. specialization in a clonal perennial (Ranunculus repens). I. Dudley, S.A. & Schmitt, J. (1996) Testing the adaptive The dynamics of ramets in contrasting habitats. Journal of plasticity hypothesis: density-dependent selection on Ecology, 69, 743–755. manipulated stem length in Impatiens capensis. American MacDonald, N. & Watkinson, A.R. (1981) Models of an Naturalist, 147, 445–465. annual plant population with seedbank. Journal of Theo- Ellison, A.M. (1987) Density-dependent dynamics of Salicor- retical Biology, 93, 643–653. nia europaea monocultures. Ecology, 68, 737–741. Newman, M.C. (1993) Regression analysis of log-transformed Fernández, C. (1999) Ecology of Sargassum muticum data: statistical bias and its correction. Environmental (Phaeophyta) on the north coast of Spain. IV. Sequence of Toxicology and Chemistry, 12, 1129–1133. colonization on a shore. Botanica Marina, 42, 553–562. Norton, T.A. (1976) Why is Sargassum muticum so invasive? Fernández, C., Gutiérrez, L.M. & Rico, J.M. (1990) Ecology British Phycological Journal, 11, 197–198. of Sargassum muticum on the north coast of Spain. Prelim- Norton, T.A. (1977) The growth and development of Sargas- inary observations. Botanica Marina, 33, 423–428. sum muticum (Yendo) Fensholt. Journal of Experimental Fletcher, R.L. & Fletcher, S.M. (1975) Studies on the recently Marine Biology and Ecology, 26, 41–53. introduced brown alga Sargassum muticum (Yendo) Fen- Norton, T.A. & Deysher, L.E. (1989) The reproductive sholt. I. Ecology and reproduction. Botanica Marina, 18, ecology of Sargassum muticum (Yendo) at different lati- 149–156. tudes. Reproduction, Genetics, and Distributions of Marine Geber, M.A. (1989) Interplay of morphology and development Organisms. Proceedings of 23rd European Marine Biology on size inequality: a Polygonum greenhouse study. Ecolo- Symposium (eds J.S. Ryland & P.A. Tyler), pp. 115–125. gical Monographs, 59, 267–288. Olsen & Olsen, Fredensborg Denmark. Givernaud, T., Cosson, J. & Givernaud-Mouradi, A. (1991) Reed, D.C. (1990) An experimental evaluation of density Étude des populations de Sargassum Muticum (Yendo) dependence in a subtidal algal population. Ecology, 71, Fensholt sur les côtes de Basse-Normandie (France). 2286–2296. Estuaries and Coasts: Spatial and Temporal Intercompari- Rice, K.J. (1990) Reproductive hierarchies in Erodium: effects © 2002 British sons. ECSA 19 Symposium (eds M. Elliot & J.P. Ducrotoy), of variation in plant density and rainfall distribution. Ecological Society, pp. 129–132. Olsen & Olsen, Fredensborg Denmark. Ecology, 71, 1316–1322. Journal of Ecology, Harper, J.L. (1977) Population Biology of Plants. Academic Salles, S., Aguilera, J. & Figueroa, F.L. (1996) Light ﬁeld in 90, 820–829 Press, London. algal canopies: changes in spectral light ratios and growth JEC_720.fm Page 829 Monday, September 23, 2002 11:06 AM

829 of Porphyra leucosticta Thur. Le Jol. Scientia Marina, 60 frequency distributions of plant mass: the absence of Density- (Suppl.), 29–38. dominance and suppression in competing monocultures of dependence in an Samson, D.A. & Werk, K. (1986) Size-dependent effects in the Festuca paradoxa. Ecology, 64, 469–475. analysis of reproductive effort in plants. American Natural- Underwood, A.J. (1997) Experiments in Ecology, Their Log- invasive seaweed ist, 127, 667–680. ical Design and Interpretation Using Analysis of Variance. Santos, R. (1995) Size structure and inequality in a com- Cambridge University Press, Cambridge. mercial stand of the seaweed Gelidium sesquipedale. Marine Viejo, R.M. & Åberg, P. (2001) The effects of density on the Ecology Progress Series, 119, 253–263. vital rates of a modular seaweed. Marine Ecology Progress Schiel, D.R. (1985) Growth, survival and reproduction of two Series, 221, 105–115. species of marine algae at different densities in natural Weiner, J. (1985) Size hierarchies in experimental populations stands. Journal of Ecology, 73, 199–217. of annual plants. Ecology, 66, 743–752. Schmid, B. & Harper, J.L. (1985) Clonal growth in grassland Weiner, J. (1990) Asymmetric competition in plant popula- perennials. I. Density and pattern-dependent competition tions. Trends in Ecology and Evolution, 5, 360–364. among plants with different growth forms. Journal of Weiner, J., Berntson, G.M. & Thomas, S.C. (1990) Competi- Ecology, 73, 793–808. tion and growth form in a woodland annual. Journal of Schmitt, J., Eccleston, J. & Ehrhardt, D.W. (1987) Dominance Ecology, 78, 459–469. and suppression in experimental radish population. Weiner, J. & Solbrig, O.T. (1984) The meaning and measure- Journal of Ecology, 75, 651–665. ment of size hierarchies in plant populations. Oecologia, 61, Schmitt, J., McCormac, A.C. & Smith, H. (1995) A test of the 334–336. adaptive plasticity hypothesis using transgenic and mutant Weiner, J. & Thomas, S.C. (1986) Size variability and com- plants disabled in phytochrome-mediated elongation petition in plant monocultures. Oikos, 47, 211–222. responses to neighbors. American Naturalist, 146, 937– Weiner, J. & Thomas, S.C. (1992) Competition and allometry 953. in three species of annual plants. Ecology, 73, 648–656. Scrosati, R. & Servière-Zaragoza, E. (2000) Ramet dynamics Winer, B.J., Brown, D.R. & Michels, K.M. (1991) Statistical for the clonal seaweed Pterocladiella capillacea (Rhodophyta, Principles in Experimental Design. McGraw-Hill, New Gelidiales): a comparison with Chondrus crispus and with Yo rk . Mazzaella cornucopiae (Gigartinales). Journal of Phycology, Yastrebov, A.B. (1996) Different types of heterogeneity and 36, 1061–1068. plant competition in monospeciﬁc stands. Oikos, 75, 89–97. Sjøtun, K., Fredriksen, S. & Rueness, J. (1998) Effect of can- Yokozawa, M. & Hara, T. (1995) Foliage proﬁle, size structure opy biomass and wave exposure on growth in Laminaria and stem diameter-plant height relationship in crowded hyperborea (Laminariaceae: Phaeophyta). European plant populations. Annals of Botany, 76, 271–285. Journal of Phycology, 33, 337–343. Yoshida, T. (1983) Japanese species of Sargassum subgenus Smith, H. (1982) Light quality, photoperception and plant Bactrophycus (Phaeophyta, Fucales). Journal of the Faculty strategy. Annual Review of Plant Physiology, 33, 481–518. of Sciences Hokkaido University Series V. Botany, 13, 99– Thomas, S.C. & Weiner, J. (1989) Growth, death and size 246. distribution change in an Impatiens pallida population. Journal of Ecology, 77, 524–536. Received 3 January 2002 Turner, M.D. & Rabinowitz, D. (1983) Factors affecting revision accepted 14 June 2002