Together but different co-occurring dune plant species differ in their

Oecologia (2014) 174:651–663

DOI 10.1007/s00442-013-2820-7

Together but different: co?occurring dune plant species differ in their water? and nitrogen?use strategies

Raimundo Bermúdez · Rubén Retuerto

received: 21 april 2013 / accepted: 28 October 2013 / Published online: 10 november 2013 © springer-Verlag Berlin heidelberg 2013

Abstract stress factors may severely constrain the range of plant physiological responses in harsh environments. convergence of traits is expected in coastal dunes because of environmental filtering imposed by severe abiotic fac-tors. however, the wide range of morphological and pheno-logical traits exhibited by coexisting dune species suggests considerable variation in functional traits. We hypothesized that the constraints imposed by structural traits ought to translate into physiological differences. Five dominant spe-cies with different morphological traits, but coexisting in a homogeneous dune area in northwest spain, were selected for study. soil characteristics and leaf functional traits were measured in april, June and november 2008. Integrated water-use efficiency (assessed by c isotope discrimination) and n acquisition and use strategies (estimated by n iso-tope composition) varied significantly among species and the differences changed over time. species differences in specific leaf area, relative water content, leaf n and c:n ratio, also varied over time. The species differed in stoma-tal density but not in soil characteristics, with the exception of ph. species differences in functional traits related to the use of resources suggest species niche segregation. spe-cies-specific temporal effects on the use of these resources support temporal niche differentiation. somewhat in con-trast to the findings of previous studies on harsh environ-ments, this study revealed a considerable level of functional

communicated by Jeremy lichstein.

r. Bermúdez and r. retuerto contributed equally to this work.r. Bermúdez · r. retuerto (*)

Department of ecology, Faculty of Biology, University

of santiago de compostela, 15782 santiago de compostela, spain

e-mail: ruben.retuerto@usc.es

diversity and complexity, suggesting that dune plant spe-cies have evolved species-specific strategies to survive by partitioning growth-limiting resources.

Keywords comparative physiology · stable isotopes · limiting similarity · niche differentiation · species coexistence

Introduction

Plants that grow in coastal sand dunes, especially fore-dune plants, are exposed to very harsh and unpredictable environmental conditions. studies in harsh environments, such as dune systems, have suggested that stress factors may severely constrain the range of plant physiological responses as result of a process of environmental filtering (reich et al. 1997, 2003; Meinzer 2003). Water and n are among the most important limiting factors for vegetation in sand dune ecosystems (Tilman 1987; sharifi et al. 1988), but other stress factors such as high temperature and light intensity, wind exposure, sand burial, salt spray and soil salinity (hesp 1991; Maun 2009) are liable to limit plant productivity and to determine community structure. The very stressful conditions imposed by the concurrence of these environmental factors, operating at variable temporal and spatial scales, coupled with the high degree of distur-bance and unpredictability of these environments (robin-son and sandgren 1983; crawford 2008), explain why only some plant species with particular adaptations, from the biochemical to the ecological level, are able to survive in such hostile environments (hesp 1991; Maun 2009).

Much research has focused on providing explanations for the maintenance of the diversity found in arid envi-ronments, such as dunes (chesson et al. 2004), with the

1 3

652functional fit between organisms and their environments being an essential feature of biological diversity (ack-erly 2004). numerous studies have examined the different mechanisms proposed to explain how competing species achieve stable coexistence (chesson 2000; angert et al. 2009; adler et al. 2013). some of these mechanisms, such as the differential use of resources (Tilman 1988) and the predation dependent on frequency (gendron 1987), are not seriously influenced by environmental variation, whereas, the action of others, such as the tradeoffs between compe-tition and colonization (hastings 1980), the relative non-linearity of competition (armstrong and Mcgehee 1980) and the storage effect (chesson 1994), depend critically of environmental fluctuations. a greater understanding of the mechanisms of coexistence in competing species makes advisable to identify species differences in functional traits associated with resource uptake (angert et al. 2009; adler et al. 2013).

Direct comparisons of physiological differences between species co-occurring in coastal dune systems are very scarce (rubio-casal et al. 2010). This is an important issue to address in order to identify differential sensibilities of species to key environmental factors. climate change is anticipated to increase the intensity and/or frequency of drought [Intergovernmental Panel on climate change (IPcc) 2004; loik et al. 2004] and dunes species are likely to be very sensitive to climate change (Maun 2009). The effects of drought are not likely to be similar for species differing in physiology and resource allocation. Thus, knowledge of species differences in physiological traits is relevant in the context of ongoing environmental changes in order to predict how habitat modifications might alter functional diversity and to impact ecosystem processes (Díaz and cabido 2001; sandquist and cordell 2007; Ber-múdez and retuerto 2013).

In this study, we applied a comparative approach to investigate some key components of the strategies used by five co-occurring foredune species to cope with the chal-lenge of surviving in such unfavourable disturbed environ-ments. The species were chosen because of their ubiquitous distribution in the coastal dune systems of northwest spain and especially on foredunes where both resource stress and disturbances by high wind, extensive sand movement and accretion, salt spray episodes and wave action are par-ticularly severe. These species, which usually co-occur in mixed communities and are likely to interact, exhibit very different life history and leaf traits (size, colour, shape, pubescence, toughness, etc.), but little is known about the ecophysiological differences between them. considering the wide range of morphological and phenological traits displayed by the species studied, it seems that survival in coastal dune environments can be achieved by differ-ent strategies resulting from the combination of different 1 3

Oecologia (2014) 174:651–663

adaptations, which suggests that there is more than one biotic solution to the same abiotic conditions. here, we hypothesized that the constrictions imposed by morpholog-ical traits ought to translate into physiological differences. To test this hypothesis, we explored whether the species differed in how efficiently they use two essential resources: water and n. specifically, we focused on the integrated water-use efficiency (WUe), assessed by c isotope dis-crimination (Δ13c) and integrated n-use and acquisition strategies, as estimated from n isotope composition (δ15n) (robinson et al. 2000). Because the enzyme rubisco reacts more readily with 12cO2 than it does with 13cO2, carboxy-lation discriminates against the heavier isotope (Farquhar et al. 1989). The extent of the discrimination depends on the concentration of cO2 at the active site of rubisco (Ci). Ci is in itself a reflection of the stomatal conductance rel-ative to photosynthetic rate. Therefore, a low Ci, and the consequent low Δ13c, may be explained either by a high photosynthetic rate or, more commonly, by a low stoma-tal conductance, which would result in a greater integrated WUe (Korner et al. 1988; Farquhar et al. 1989). δ15n has been proposed as an integrative trait to assess stress toler-ance in plants (robinson et al. 2000), although there is not yet a clear interpretation of low vs. high δ15n values (see “Discussion” for details). Many factors contribute to the variations in 15n abundances of plants, among them, iso-topic fractionation (during decomposition and n uptake, and fractionation within plants) and differences in plant growth and resource-acquisition strategies (source of plant n, depth from which n is acquired, presence of mycor-rhizal associations, chemical form of the n taken up and seasonality of n uptake) (McKane et al. 1990; nadelhof-fer et al. 1996; Filella and Peñuelas 2003). Therefore, from the study of natural variations in 15n abundance it is pos-sible to infer the spatiotemporal partitioning of n, a lim-iting resource for dune plant growth and consequently an important determinant of species composition and commu-nity structure. Because soil quality can show variation over only a few centimetres (salzman 1985; lechowicz and Bell 1991) we analysed characteristics of the soil in which spe-cies were rooted. In addition, we measured a number of relevant leaf traits related to light, water, and n use, such as c, n and water content, specific leaf area (sla; leaf area/leaf mass), which is related to photosynthetic capac-ity, leaf longevity and stress tolerance (Westoby 1999), and stomatal density (sD), which contributes to stomatal con-ductance, which in turn regulates the diffusion of cOthe leaf during photosynthesis and the outward diffusion 2 into of water vapour (hetherington and Woodward 2003). To increase our understanding of Δ13c variation, we examined relations between Δ13c and leaf traits related to photosyn-thetic rate, such as n content and sla, and to stomatal conductance, such as water content, the two processes that

Oecologia (2014) 174:651–663 contribute to explain Δ13c. Differences in these variables and/or relations may contribute to these species having species-specific patterns of resource use, which may play a crucial role in the segregation of their ecological niches and therefore in explaining stable coexistence of species.

Materials and methodsspecies and site

Five co-occurring perennial herbaceous species, which are among the most abundant in the coastal dunes of galicia (northwest spain), were selected for study. These species belong to five different families and display very differ-ent leaf characteristics. They were the most abundant in the community studied in terms of biomass and numbers. Eryngium maritimum l. (Umbelliferae) is a hemicrypto-phyte with three-lobed folded leaves. The plant has stiff, spiny leaves, which have a leathery texture and are bluish-green in colour. Euphorbia paralias l. (euphorbiaceae) is a chamaephyte with fertile and sterile branches and closely overlapping leaves. The leaves are of mean size 20 mm and are ovate to elliptical in shape. Pancratium maritimum l. (amaryllidaceae) is a geophyte (bulbous plant). a long stem grows from the bulb and ends in linear-shaped glau-cous leaves. Matthiola sinuata (l.) r. Br. (Brassicaceae) is a hemicryptophyte that grows as a basal rosette. The leaves and stems are covered with thick grey down. Calys-tegia soldanella (l.) r. Br. (convolvulaceae) is a hemic-ryptophyte with fleshy, kidney- or heart-shaped leaves. It also has large trumpet-shaped, pink to purplish pink flow-ers with lighter stripes. all these species have clism, although C. soldanella and E. paralias can switch to 3 metabo-c4 metabolism and E. paralias also to crassulacean acid metabolism (caM), depending on habitat factors (elhaak et al. 1997; Daniela et al. 2009). all selected species are amphistomatic, except E. paralias, which is epistomatic (Daniela et al. 2009).

In the coastal dune at lariño (galicia: 42°46′00″n, 9°06′58″W), we delimited an area of 250 m2 in the fore-dune. This area can be considered homogeneous at cer-tain scales [micro-β scales, sensu Orlanski (1975)]. climatic data (summarized in Fig. 1) were obtained from the nearest meteorological station at corrubedo (42°33′20″n, 9°01′43″W; 25 km away from lariño) and were provided by the meteorological service of galicia (http://www.meteogalicia.es). The disturbance regime at the site, away of large cities, is mainly natural, due to heavy storms during winter time. Within this area, 12 dif-ferent individuals of each species were randomly sampled in april, June and november 2008. In april, all species (hemicryptophytes, chamaephytes and geophytes) had

653

Fig. 1 gaussen’s ombrothermic diagram for the period 2000–2013 obtained from the meteorological station at corrubedo (42°33′20″n, 9°01′43″W). Monthly average temperature (°c) and monthly average precipitation (mm) in a 2:1 (temperature:precipitation) scale. Drought conditions are considered to occur when the precipitation is lower than twice the temperature value

resumed their growth after winter and showed new, fully expanded leaves. June represents a critic period for plant life due to high temperatures, radiation and drought, and it is the only month in which all species may coincide in the flowering period. november represents the end of the veg-etative period for the species studied, with leaves showing signs of senescence.

leaf relative water content, sD and sla

Plant water status was calculated from the relative water content (rWc) of one fully expanded leaf per plant (the mean value for five leaves in E. paralias). The leaves were excised in the field and transported to the laboratory in sealed zip-lock, tared plastic bags. a whole leaf of M. sinuata and C. soldanella, a distal piece of leaf (ca. 8 cm) of P. maritimum, and one disc (17 mm in diameter) taken from the distal part of a leaf of E. maritimum were used for measurements. Because of the small size of E. paralias leaves, the rWc was measured on whole leaves. The use of different procedures in sampling leaves for rWc was unavoidable given the wide range of variation in the leaf size of the species. The fresh mass (Mals was determined. Turgid mass (Mf) of these materi-t) was measured after placing leaf material on filter paper saturated with distilled water in covered Petri dishes for 20 h at 4 °c in the dark. The leaf material was blotted dry with tissue paper before the Mt was measured. The dry mass (Md) was determined after oven drying at 75 °c to constant weight. The rWc was calculated as [(Mf ? M).

d)/(Mt ? Md)] × 100 (Peñuelas et al. 2004stomatal densities (no. stomata cm?2) were determined

on one leaf per plant from impressions of the abaxial and

1 3

654adaxial mid-portions of leaf surfaces (in E. paralias, sto-mata are absent from the adaxial surface), which were obtained by applying a thin layer of clear nail polish. The dried nail-polish impressions were peeled from the leaves and mounted on a slide, and the stomata were counted in 100× micrographs. One leaf was used per plant and three micrographs were captured per leaf (three leaves and one micrograph per leaf for E. paralias). Two 500 × 500 μm frames per micrograph were used to calculate the sD. Pol-ish imprints could not be obtained from M. sinuata because of the dense pubescence on the leaf surface.

sla was calculated for each individual plant on one fresh, fully expanded leaf of M. sinuata and C. soldanella, a distal piece of leaf (ca. 8 cm length) for P. maritimum, one leaf disc (17 mm in diameter) for E. maritimum and the average of five leaves for E. paralias. The image of the projected leaf area was determined with a flatbed scanner (canoscan liDe 50; canon, Tokyo) and processed with a freely available public-domain image processing and anal-ysis program [ImageJ software; national Institute of health (rasband 1997)]. The leaf material was then oven-dried at 75 °c to constant weight and the mass was determined. The sla was calculated as the one sided projected area of foliage per unit dry mass (cm2 g?1 dry mass) (cornelissen et al. 2003).

leaf c and n content and isotopic composition

at all three sampling times, several leaves from each selected plant were cleaned of organic debris, dried at 75 °c for over 5 days and then ball-milled to a homog-enized fine powder to pass through a 40-μm-mesh screen. The concentrations of total c and n (percentage of dry mass) and the molar 15n/14n (δ15n) and 13c/12c (δ13c) ratios were determined on subsamples (ca. 1–2 mg dry wt), in an elemental analyser (Flashea 1112 series) cou-pled to an isotope ratio mass spectrometer (MaT253; ThermoFinnigan, Bremen, germany). c and n isotope ratios were expressed relative to the composition of a standard [Pee Dee belemnite (PDB) cacOatmospheric n for n]. The δ values (‰) were calculated 3 for c, and as [(Ror 15n/sam/R14n ratio in the plant sample and standard. Polyeth-std) ? 1] × 1,000, where R refers to the 13c/12c ylene [International atomic energy agency (Iaea)-c6)] and (nhnational isotope standards for c and n, respectively. The 4)2sO4 (Iaea n1) were used as secondary inter-δ13c values were transformed into Δ13c values by use of the following expression: Δ13c = (δ13cair ? δ13c(1 + δ13cplant)/plant) (Farquhar et al. 1989), assuming a δ13c air value of ?8 ‰ on the PDB scale (Farquhar et al. 1989). In c3 plants, Δ13c values generally range between 15 ‰, for plants with higher WUe, and 25 ‰, for plants with lower WUe (griffiths 1992).

1 3

Oecologia (2014) 174:651–663

soil measurements

In June 2008, the characteristics of the soil in which spe-cies rooted were measured. One soil sample (ca. 25 cm in depth, 5 cm diameter) was collected at a distance of 10 cm from the base of six randomly selected plants of each spe-cies, and another six soil samples were collected from bare soil. The soil samples were transported in a cool box (4 °c) to the laboratory, in hermetically sealed containers, and immediately weighed upon arrival. The samples were then dried to constant weight at 75 °c and weighed, in order to estimate the soil water content (g water kg?1 dry soil). sub-samples of soils were used to determine the amount of inor-ganic P available to plants, by the Olsen’s method (Olsen et al. 1954). Dried subsamples (approximately 2 mg) of each sample were ground, sieved to <2 mm and encapsu-lated in tin capsules. The samples were then analysed for c, n content via dry combustion at high temperature, in a chns elemental analyser (lecO chns-932; lecO, st. Joseph, MI). soil salinity was estimated as electrical con-ductivity, measured in a soil: water suspension (ratio 1:2.5), with an electrical conductivity meter (524 crison; crison Instruments, Barcelona). Further subsamples (from a 1:2.5 dilution) were used to measure the ph in water and in Kcl, with a ph-meter (ph-metro 507; crison Instruments).statistical analysis

Differences in leaf traits, %c, %n, Δ13c, δ15n, sla, rWc and sD, were analysed by two-way anOVa, with r sta-tistical software (r Development core Team 2009). spe-cies and sampling date were considered as fixed factors. The differences in soil samples for the variables considered were determined by one-way anOVa. Prior to analyses, all variables were checked for normality and homoscedasticity required for anOVa. The c:n ratio and sD were log trans-formed prior to the analysis so that they met the underlying assumptions of the test. The slight deviations found for the other variables warranted the use of anOVas because this method is sufficiently robust to perform well when data do not deviate greatly from the underlying assumptions (Zar 1984). Post hoc pairwise comparisons between means were carried out by using Tukey honest significant difference (hsD) tests [TukeyhsD function of the package stats of r (r Development core Team 2009)]. significance level for all tests was set at α = 0.05.

We used the r statistical software lm function from the stats package (r Development core Team 2009) to test for association between Δ13c and %n, δ15n, sla and rWc. a rotated principal component analysis (Pca) based on a correlation matrix of species by six leaf traits was per-formed with the aim of summarizing the main factors determining most of the variation in the data. The prcomp

Oecologia (2014) 174:651–663

function of the stats package (r Development core Team 2009) was used to perform the Pca.

Results

leaf functional traitsLeaf N content

species differed significantly in their leaf n concentra-tions, and the extent of these differences changed signifi-cantly over time (Table 1). There was a general trend in all species for n concentrations to be lower in June than in april, when the n concentration was highest (P always <0.001; Fig. 2a), with intermediate values in november. In E. paralias, C. soldanella and P. maritimum the n con-centrations had fully recovered in november, but not in M. sinuata and E. maritimum. at all times, the n concentration was highest in P. maritimum and lowest in E. paralias.Foliar C:N ratio

Differences in c:n ratios of species varied significantly between sampling dates (Table 1; Fig. 2b). The c:n ratios in E. paralias, in which the c:n ratio was highest on all dates, and in M. sinuata and E. maritimum, were clearly higher in June than in april. In the two first species, the c:n ratios in november almost recovered to those initially observed in april, in E. maritimum the values were similar in June and november. The c:n ratios in P. maritimum and C. soldanella remained fairly constant over time.Leaf N isotope composition δ15N

species differences in leaf δ15n varied significantly between the sampling dates (Table 1). There was no general pattern in the variation in δ15n values of species between the sampling dates (Fig. 2c). The δ15n values in M. sinuata and P. maritimum tended to decrease slightly between april and november, and in E. paralias and E. maritimum these values decreased significantly between april and June and then in november recovered or clearly exceeded (E. mar-itimum) the values observed in april. C. soldanella was the only species in which the δ15n values constantly increased between april and november. The δ15n values were high-est in M. sinuata and lowest in E. paralias for all dates throughout the period of study.C isotope discrimination

according to Δ13c values, species-related differences in WUe varied significantly between the different sampling

no*i212

t638a...n010iF

9mirc2s97

id6222, 0671cis

,s323,,9p1125otosiD

326

3 csf

2d1 ,)N***07151240...δ(F556 41noit090is6721os1508ps,,,8355moc

c4288 cirWf

4d1poto**s32*i529 n771,, F860,,)626N:C0205(4396 3533ois,,,,tas91025113r na

:cl4280 sf

5d1,)N***%437(777 ...tnF481421etnosc s03797235n[c

3s14280tiΔf

5d1art f***a546e151l... laF5534n]

o)itD8178c3215nS(....u fys90632247 tins32son een

td54280a l1f

5daδd1 tdanm**a74* otss019e 703,,i,c)575,eCpF568211sW fRo( 8265 t3778sntces0988tens,,61ffoec n

4280ere:htcf

5d1ta row fe*** v483it034a...l568OVaenr F2111,) aAl3066aL2523iSr(s....o s1888621tacearfa n

4280

ffoa%f

5d1 els tlcufietsiaeced rps×5)

s

s1 , l0.(ssa ) eee0ru lCoiib3ccd1ti≤ceetes TaΔapaeP(FsDpsr *655

1 3