The role of coastal ecotones: a case study of the salt marsh/upland transition zone in California

Authors

  • BIBIT HALLIDAY TRAUT

    Corresponding author
    1. Wildlife, Fish, and Conservation Biology, University of California, One Shields Avenue, Davis, CA 95616, USA
    Search for more papers by this author

Present address and correspondence: Bibit Halliday Traut, Department of Environmental Science, Policy and Management, Ecosystem Sciences Division, 151 Hilgard Hall #3110, University of California, Berkeley, CA 94720, USA (e-mail btraut@nature.berkeley.edu).

Summary

  • 1Ecotones, the transition zones between adjacent ecological systems, may intensify or concentrate the flow and processing of materials and organisms between systems.
  • 2To determine whether the salt marsh/upland transition zone is an ecotone, both in functioning as a habitat, as well as concentrating materials and organisms, I conducted an extensive survey of 12 marsh/upland transition zones in the Pt Reyes area of California.
  • 3The high saltmarsh was identified as an ecotone, with biological and physical conditions distinct from the adjacent marsh plain and upland. It supports a unique plant assemblage and greater total soil nitrogen than the adjacent upland.
  • 4Taxonomic richness patterns did not reflect changes in habitat spatial characteristics (e.g. ecotone area, ecotone area/perimeter), instead showing a quadratic correlation along the gradient, with increased diversity of both plants and spiders in the ecotone. Spider richness was enhanced with increased vegetation complexity, which was greatest in the ecotone.
  • 5This study presents empirical evidence that suggests coastal marsh/upland ecotones harbour increased diversity and may concentrate nitrogen pools, thereby highlighting the need for further research investigating the relationship between landscape connections and the coupling of species and biogeochemical processes.

Introduction

Understanding the transition between the components of a landscape can help increase effective management of these systems and the role of ecotones is therefore receiving renewed interest and support (Risser 1995; Pollock et al. 1998; Levin et al. 2001). The characteristics of these transition zones are uniquely defined by different scales of space and time and by the strength of interactions between the adjacent ecosystems (Holland 1988) but ecotones may share compositional and structural characteristics with one or other of the habitats they border, in addition to having their own unique physical and biogeochemical characteristics (Johnston 1993; Risser 1995; James & Zedler 2000). Such transitional zones may be diversity hotspots (e.g. passerines and ants, Kark et al. 2002; wetland plants, Kirkman et al. 1998), and their spatial characteristics of the ecotone (such as area and perimeter) may reflect changes in species richness patterns (Helzer & Jelinski 1999). Ecotones frequently intensify or concentrate the flow and processing of materials (Wiens et al. 1985; Johnston 1993; Mitsch & Gosselink 1993) and nutrient retention (e.g. of nitrogen) may also be related to their spatial configuration (length vs. area; Holland et al. 1990). Wetlands, in particular, have been shown to be important areas of nitrogen retention (Johnston 1993; Woltemade 2000; Bai et al. 2004).

Salt marshes, which occur along coastlines at mid to high latitudes (Mitsch & Gosselink 1993), are relatively narrow zones that provide regulate fluxes of nutrients, water and organisms between ecosystems and also play a key role in storing and removing nutrients and pollutants (Risser 1995; Levin et al. 2001). In Californian salt marshes, the border between the marsh and upland is marked by a narrow zone (the high salt marsh; Zedler et al. 1999) where abrupt changes occur in plant physiognomy and species composition and this marine/terrestrial transition zone is clearly an ecotone where attributes of both landscape types are overlain by a steep gradient in physical variables that directly affect the distribution of organisms and ecological processes (Levin et al. 2001).

Although the high salt marsh transition zone is at the node of overlap of several species’ distributions, most of the work to date has focused on the zonation pattern across this marine/terrestrial transition and not on the ecotone as a habitat itself. For example, the model that is most often used to describe the tidal marsh zonation pattern (e.g. Pennings & Callaway 1992; Sanchez et al. 1996) depends on elevation, which may be related to the soil edaphic conditions (e.g. salinity, redox, moisture, texture) that are likely to determine species distributions (Cantero et al. 1998). Zedler et al. (1999), however, suggest that models should also include landscape position and conspicuous species. Along the steep environmental gradients of salt marshes, competitive relationships between plants can be affected by nitrogen supply (Valiela & Teal 1974; Levine et al. 1998) and deposition of dung by herbivores may lead to increased mineralization (Holland & Dettling 1990). Grazing can also change plant biomass and structure (Kiehl et al. 1996; Ford & Grace 1998) and alter competitive relationships between salt marsh plants (Bos et al. 2002) and therefore affect invertebrate communities (Andresen et al. 1990).

Salt marshes, a naturally restricted habitat important for conservation worldwide (Adam 2002) are threatened by a dramatic decline in area and water quality. In California, where the present study took place, approximately 85% of the original tidal marshes have been destroyed or degraded (Dedrick 1989). Efforts to protect the remaining marshes and restore degraded marshes are being made although there is little understanding of the dynamics of the system. The lower tidal areas of California salt marshes (dominated by Spartina foliosa and Salicornia virginica) have received attention, but few studies have documented species distributions across Pacific coast wetland-upland ecotones (for an exception see James & Zedler 2000) or related their unique structure to other trophic levels (e.g. generalist predators such as spiders).

I investigated whether a distinctive community of plants characterizes the high salt marsh at several sites in the Point Reyes peninsula of California, and which indicator species and environmental conditions best describe it. I also tested whether this transition zone concentrates materials (nitrogen) and organisms (plant spiders), as other ecotones have been shown to do.

Materials and methods

study area

This study was conducted in coastal salt marshes in the Point Reyes area (38° N, 122° E) of northern California incorporating Point Reyes National Seashore and adjacent protected areas. Several relatively pristine salt marshes, ranging in size from 11 km2 to over 235 km2, with intact high marsh habitats (Evens 1993), persist within varying contexts of land use and form (e.g. along the tectonic border of the Pacific Plate and North American Plate). Although physically separated, the high marsh areas are linked by dispersal of propagules via water currents, wind and animals. The low marsh is typically dominated by Spartina foliosa and the marsh plain by Salicornia virginica and Distichlis spicata, whereas the species-rich high marsh includes Salicornia virginica, Distichlis spicata, Frankenia salina, Grindelia stricta, Jaumea carnosa, Limonium californicum and Triglochin spp. (Barbour & Major 1988; Zedler et al. 1999). Plant nomenclature follows Hickman (1996). Spiders are common inhabitants of salt marshes, seen on the marsh surface and vegetation and often spotted swimming on the incoming and outgoing tide.

vegetation sampling

In June and July 1999, I conducted an inventory of 12 salt marshes within the Point Reyes area. Aerial photographs (provided by Point Reyes National Seashore) were used to identify high marsh areas (a distinct band between the marsh plain and upland where both canopy surface heterogeneity and the presence of woody plants increase) and to measure their area and perimeter. These aerial photographs were also used to determine the area and perimeter (using the ArcView Geographic Information System, Environmental System Research Institute Inc., Redlands, CA, USA). The marked change in physiognomy between ecosystems is a characteristic often used to identify ecotones (Walker et al. 2003).

I focus on the sharp transition between the marsh plain and upland, an area little understood in the relatively spatially restricted Pacific coast salt marshes. In other salt marsh ecosystems, such as those along the east and gulf coasts of N. America, the term high salt marsh is often used for a broader zone, from the marsh plain down to the MHW mark (Bertness & Ellison 1987), and its upland border is probably more comparable with the ecotone studied here.

At each of the 12 marshes, 15 transects were placed randomly across the high marsh perpendicular to the water's edge. In most cases, transects extended 1 m on either side into the upland and the marsh plain, but some were located in high marsh areas found on elevated islands (patches) within the marsh and along creeks with natural levees, in order to characterize the relationship of the high marsh vegetation to environmental factors. Transects ranged in length from 3 m to 83 m. Quadrats (0.5 m long × 1 m wide, n = 1226) were placed with their long edge perpendicular to the transect, at 2-m intervals starting at a random point within the marsh plain, and vegetation cover (as modified Daubenmire cover classes; Muir & McCune 1988) was recorded. Transect and quadrat locations were recorded with a Geographic Positioning System (Trimble Pro-XR).

environmental factors

A small soil core (2 cm3 from a depth of 5 cm) was collected from the centre of most quadrats (n = 804 of 1226), to determine salinity. Cores were not collected during a specific stage in the tidal cycle, but were collected throughout the day as vegetation was sampled. Logistical constraints dictated that only a subset could be sampled and quadrats in those sections of the transect that passed across homogenous vegetation were selected. Saturated soil pastes (Richards 1954) were made from the soil cores and a drop of water was extracted through filter paper onto a temperature-compensated refractometer to measure salinity (±1.0). Steel rods were sunk into the substrate (> 50 cm depth) of each quadrat sampled for salinity and those that could be relocated at the end of the main growing season (n = 765) had the depth of the rust line measured. This estimated the depth from the soil surface to the anoxic zone, where rust can no longer form. Elevation above MHW (the transition between Salicornia virginica and Spartina foliosa: relative elevation = 0.0 m) was also measured at the centre of each ‘salinity’ quadrat with laser controlled survey equipment. In a subset of three transects in each marsh, three additional soil cores (3 cm diameter × 20 cm depth) were randomly taken from each quadrat to measure substrate water content, total Kjeldahl nitrogen (TKN) and texture (percentage clay, silt and sand). The same volume of soil from each sample was weighed before and after drying to constant weight (60 °C for 48 hours) and water content calculated as a percentage of dry soil weight lost (n = 265). A subsample of the dried soils (n = 196) was then analysed for TKN and texture by the DANR Soil Analytical Laboratory, University of California, Davis.

parallel transect sampling

In June and July 2000, I conducted a survey in four marshes to determine patterns of plant family and spider family richness across the high marsh. Spiders have not been well studied in these systems and are difficult to identify to genus/species and family diversity accurately predicts species richness for both vascular plants and invertebrates (Whitmore et al. 2002; Báldi 2003; Sauberer et al. 2004). In each marsh, two transects (2 m × 10 m) were randomly placed parallel to and centred in the high salt marsh (defined by the presence of Grindelia stricta, identified as a high marsh indicator species from results of vegetation sampling of natural, ungrazed, high marshes), and matched by transects in the adjacent marsh plain and upland. All plant species occurring in each transect were recorded and vegetation height measured every 0.5 m along the transect. Vegetation tip height diversity (VTHD) was calculated as the Inverse Simpson Index where inline image where pi is the proportion of tip height in the ith class (following methods used by Greenstone 1984). Spiders were collected by D-VAC sweeping at 1-m intervals (2 m wide) along the transect (this collects c. 97% spiders from low-profile vegetation, see Döbel et al. 1990) and in five randomly assigned, evenly spaced pit traps (2.5 cm diameter × 12.7 cm long, filled with the preservative propylene glycol) left in place for 48 hours during a neap tide so as to avoid flooding by tidal inundation. All spiders collected were identified to family using nomenclature following Ubick (1999).

analysis

Analyses were carried out using the statistics programs SPSS 9.0 (SPSS Inc. 1999) and PC-ORD (McCune & Mefford 1999). In order to investigate how plant species were distributed along the high salt marsh, I analysed the vegetation data only from quadrats on transects that extended across the ecotone (n = 855 quadrats) using non-metric multidimensional scaling ordination (NMS; Kruskal 1964) with a Sørenson distance measure. I used the multi-response permutation procedure (MRPP) to test for differences between plant community composition in the marsh plain, ecotone and upland zones, as it does not require multivariate normality and homogeneity of variances (McCune & Mefford 1999). Tests of significance are based on an approximated P-value from a Pearson Type III distribution (McCune & Mefford 1999). The Effect size (A) is a descriptor of within-group homogeneity, compared with the random expectation, and ranges between 0 and 1. For community data, A is commonly below 0.1 and values above 0.3 are considered high (McCune & Mefford 1999).

In order to identify predictors of the ecotone, I examined correlations of environmental factors with ordination axes, using Pearson's r and overlays of environmental variables on ordination diagrams. This technique does not have the problems associated with multiple regressions of community gradients on environmental variables (as in canonical correspondence analysis), nor does it assume that the environmental variables are related to the community gradient (McCune & Mefford 1999). Joint plots of overlays were constructed showing only environmental variables with r-squared = 0.03. The angle and length of the variable line indicate the direction of the strength and relationship (McCune & Mefford 1999). Additionally, general environmental characteristics for each zone were described by mean and standard error and the Kruskal–Wallis paired sample test was used to test for differences between communities.

Indicator species analysis (Dufrene & Legendre's 1997 method; PC-ORD 1999), which combines information on species abundance and occurrence of species in particular groups to calculate an indicator value, was applied to data from transects that extended across the ecotone. Statistical significance is determined by a Monte Carlo technique. Because the 12 marshes may have somewhat different environmental characteristics, I also performed the indicator species analysis for each marsh. Data from all transects (across ecotone, over patches and along creeks) were used in linear regressions examining the relationship between indicator species’ distributions and environmental variables.

Using all salt marsh sites (n = 12), I used linear regression to examine the relationship between plant richness and high marsh area and high marsh area/perimeter (a measure of edge). Quadratic regression analysis of the parallel transect samples (where transects lay within, rather than across, zones), was used to determine if an increase in spider and plant family richness was found within the ecotone.

Non-parametric Kruskal–Wallis tests were used to examine differences in nitrogen pools (TKN) and the relationship between TKN and environmental variables was examined using non-parametric Spearman correlations.

Results

structure of the high salt marsh

I found a total of 121 plant species in the 12 Point Reyes marshes. The high salt marsh plant assemblage is distinct from the marsh plain and the upland communities (Fig. 1) and variation is predominately accounted for by the first axis of the NMDS ordination (Fig. 1; axis 1, R2 = 0.576; axis 2, R2 = 0.175). Generally, assemblages within the upland community were found to the far right of axis 1, and marsh plain assemblages to the far left, whereas the high marsh community was spread out across the gradients of both axes 1 and 2, but was restricted from the extremes of either axis. Plots found in the high marsh were significantly different in terms of species composition and abundance from those found in the marsh plain and upland (MRPP, P < 0.001, effect size (A) = 0.16). Marshes with greater area or edge (smaller high marsh area/perimeter) did not have significantly different native or exotic plant richness (R2 < 0.07, P > 0.05).

Figure 1.

Non-metric multidimensional scaling ordination of plots (n = 855) from ecotone transects in plant species (n = 121) space. Marsh plain plots are represented by ○, the high marsh is by γ and the upland by Δ. Plots were significantly different among zones based on species composition (MRPP, P < 0.001).

Overlays of environmental variables on ordinations of plots in species space revealed important gradients relating three environmental variables, salinity, depth to anoxia and elevation, to plant communities (Fig. 2). Variation in this subsample of plots (n = 741 vs. 855 in Fig. 1) was again predominately accounted for by the first axis (Fig. 2; axis 1, R2 = 0.553; axis 2, R2 = 0.178). Salinity and elevation were somewhat more strongly correlated to this axis (r = −0.552 and 0.592, respectively) than was depth to anoxia (r = 0.393). Generally, plots in the high salt marsh were more moderate in all three variables, with marsh plain plots having high salinity, low elevation plots with the anoxic zone close to the surface and higher, upland plots being less saline with deeper oxic zones.

Figure 2.

Non-metric multidimensional scaling ordination of plots (n = 741) from ecotone transects in plant species (n = 115). Only plots where salinity, depth to anoxia and elevation data were available were included. Overlays represent environmental variables that were significantly correlated (r-squared > 0.2) with ordination axes. Ecotone area and ecotone area/perimeter were not correlated to ordination axes and, therefore, were not included. Plot symbols as in Fig. 1.

Mean values for salinity, TKN, elevation, percentage of water and depth to anoxia differed significantly across zones (Fig. 3), but soil texture did not.

Figure 3.

Mean values (± 1 SE) for environmental variables for the marsh plain (mid marsh), high marsh (ecotone) and upland along transects crossing the high salt marsh ecotone. Salinity, elevation, depth to anoxia and percentage water were significantly different between each zone (Kruskall Wallis test, P < 0.05), but TKN was only significantly different between the upland and the other zones.

Species correlations to ordination axes are not reported here because many showed hump-shaped distributions with greater cover in plots in the middle of the ordination than on either end of the axes (see Traut 2003). Indicator species analysis nevertheless showed eight species to be significant indicators of the high salt marsh across the entire region, although each individual marsh was characterized by a different subset of these (Table 1). The indicator value was generally higher for particular species in individual marshes than for the entire region, probably due to smaller species pools and less variation. Marshes differed in environmental conditions, as indicated by physicochemical variables (see Traut 2003).

Table 1.  Results from indicator species analysis for overall region (includes all marshes) and for each local marsh. Abbreviations for sites are: HB (Home Bay), IB (Indian Beach), JO (Johnson's Oyster), KI (Kent Island), LS (Limantour Spit), MP (Millerton Point), MS (Millerton Sidepocket), PG (Pine Gulch), SB (Schooner Bay), TO (Tomales Oyster), WC (Walker Creek), and WG (White Gulch). Species abbreviations are: Ds (Distichlis spicata), Fs (Frankenia salina), Gs (Grindelia stricta), Jc (Jaumea carnosa), Lc (Limonium californicum), Pi (Parapholis incurva), Po (Polypogon monspeliensis), and Tm (Triglochin maritima). An asterisk designates an exotic species. RA represents relative abundance in the high marsh calculated by the average abundance of a given species in a given group (e.g. high marsh) of plots over the average abundance of that species in all plots in all zones (marsh plain, high marsh, and upland). RF represents percentage of plots in high marsh where given species is present. IV (indicator value) represents percentage of perfect indication of species for indicating the ecotone. Only species with indicator values ≥ 10 and with P ≤ 0.05 are reported
Site DsFsGsJcLcPi*Po*Tm
RegionRA5271795670946578
RF9034205831102321
IV4624163222101516
HBRA581007891
RF100255044
IV58253940
IBRA5181
RF10056
IV5245     
JORA59
RF89
IV52
KIRA505610058100100
RF979526843229
IV485326493229
LSRA7610072
RF515851
IV395837
MPRA8973
RF5452
IV4838
MSRA6610082
RF824978
IV544964
PGRA52
RF78
IV41
SBRA645586
RF946939
IV613833
TORA6991
RF6433
IV4430
WCRA47856981
RF94497651
IV44425241
WGRA67
RF75
IV50

Regressions of cover values revealed that individual species responded to the environmental gradients that had been detected in ordination overlays, albeit with different patterns (Table 2). For example, ungrazed marshes with low depth to anoxia (e.g. Limantour Spit and Walker Creek) generally had Grindelia stricta as an indicator species, whereas those where anoxic soils were deeper (Millerton Point and Millerton Side) had Triglochin maritima and Limonium californicum (Tables 1 and 2). In grazed marshes (Home Bay and Schooner Bay), Distichlis spicata was an indicator of the high marsh, particularly in plots with high soil water content, whereas the exotic Polypogon monspeliensis was typical at lower salinity with more freshwater influence (Tables 1 and 2). Decreased clay in the soil (e.g. Home Bay, Indian Beach and Kent Island) favoured Frankenia salina (Tables 1 and 2). Several marshes with the deeper anoxic zones were characterized by Jaumea carnosa if wet, or by Parapholis incurva where soils were drier and sandy and had less nitrogen.

Table 2.  Regression coefficients of relationship between indicator species (Spp.) for the high salt marsh and environmental variables
Spp.Salinity () (n = 804)Anoxia depth (cm) (n = 765)TKN (%) (n = 196)Sand (%) (n = 196)Silt (%) (n = 196)Clay (%) (n = 196)Water (%) (n = 265)Elevation (cm) (n = 807)Grazing (n = 1226)
  • *

    P < 0.05,

  • **

    P < 0.01.

  • Positive relationship indicates that grazing was related to increased species cover.

Ds 0.133**−0.137** 0.122−0.174* 0.128 0.242** 0.245**−0.321** 0.090**
Fs−0.053 0.205**−0.131 0.237**−0.194*−0.278**−0.145*−0.107**−0.223**
Gs−0.203** 0.246**−0.084 0.005 0.037−0.108−0.154* 0.008−0.237**
Jc 0.334**−0.268** 0.192*−0.253** 0.218** 0.270** 0.427**−0.344**−0.013
Lc 0.126**−0.075*−0.139*−0.022 0.045−0.040 0.061−0.196**−0.280**
Pi−0.161** 0.157**−0.226** 0.289**−0.262**−0.276**−0.223** 0.071*−0.062*
Pm−0.155** 0.095*−0.098−0.081 0.150*−0.116−0.114 0.238** 0.038
Tm 0.201**−0.134** 0.090−0.059 0.066 0.025 0.252**−0.016−0.151**

function of the high salt marsh ecotone

Total soil nitrogen (TKN) was higher in the ecotone than in the adjacent upland (P = 0.04), but it was not significantly different from TKN in the marsh plain (Fig. 3). Non-parametric Spearman's correlation analysis showed that TKN was positively related to wetter soils that were more saline and fine-textured (clay and silt), and negatively correlated to sandy soils with deeper anoxic zones (P < 0.01).

Richness of spider and plant families along parallel transects generally followed a quadratic relationship, with increased richness within the ecotone and decreased richness in the marsh plain and upland areas (Fig. 4a,b). Vegetation tip height diversity was greatest in the high marsh ecotone and least in the marsh plain (Fig. 4c). Twenty-eight plant families were recorded, with many having a higher relative frequency in a particular zone (Table 3).

Figure 4.

Relationship between zone location (marsh plain (mid marsh), ecotone) and (a) plant family richness, (b) spider family richness, and (c) vegetation tip height diversity richness (n = 8 for each zone). Plant richness, spider richness and vegetaion tip height diveristy showed a significant quadratic relationship to zone locaton, with increased richness in the ecotone (respectively r2 = 0.42, 0.43 and 0.63).

Table 3.  Plant families observed within transects sampled parallel to and within each zone (marsh plain, high marsh, upland)
FamilyNumber of transects where present (n = 24)Relative frequency (percentage of transects in given zone where present)
Marsh plainHigh marshUpland
Aizoaceae 5  0 3825
Apiaceae 5  0 1350
Asteraceae20 6310088
Betulaceae 2  0  025
Boraginaceae 2  0  025
Brassicaceae 3  0 1325
Caryophyllaceae 4 25 25 0
Chenopodiaceae15100 88 0
Convulvulaceae 2  0 25 0
Cuscutaceae 6 25 50 0
Cyperaceae 4  0 2525
Fabaceae 6  0 1363
Frankeniaceae13 3810025
Gentianaceae 3  0 1325
Geraniaceae 2  0 25 0
Iridaceae 2  0  025
Juncaceae18 75 7575
Pinaceae 2  0  025
Plantanaceae 2  0  025
Plumbaginaceae 8 50 50 0
Poaceae2310010088
Polygonaceae 7  0 1375
Primulaceae 6  0 2550
Rosaceae 7  0 6325
Salicaceae 1  0  013
Scrophulariaceae 7 50 38 0
Solanaceae 3  0  038
Urticaceae 1  0  013

Nineteen spider families were recorded within the marsh plain, high marsh and upland and counts enabled indicator values for each zone to be calculated (Table 4). The marsh plain was most strongly indicated by the hunting spiders, Lycosidae (wolf spiders). The large range of spiders in the ecotone had a variety of feeding types and sizes, with Araneidae (orb weavers) and Leptonetidae (leaf litter spiders) having the highest indicator values. In the upland, small web spiders like Theridiidae (cobweb weavers) and Linyphiidae (sheet-web weavers) were the strongest indicators. Linear regression showed a positive relationship between increased vegetation tip height diversity and spider family richness (R2 = 0.24, P < 0.02).

Table 4.  Spider families observed and counted within transects sampled parallel to and within each zone (marsh plain, high marsh, upland). For each family, the relative frequency (RF, represents percentage of plots in zone where given species is present) and indicator value (IV, represents percentage of perfect indication of species for indicating the zone, based on a calculation combining RF and relative abundance) are reported
FamilyToal collected (n = 1998)Marsh plainHigh marshUpland
IVRFIVRFIVRF
  • *

    Hunting spiders.

Agelenidae  1131300  0 0
Amaurobiidae*  10000 1313
Anyphaenidae* 12003822 2510
Araneidae13438210066 6318
Clubionidae* 696323501210041
Corinnidae  7134257 2511
Cybaeidae 2450336313 13 2
Dictynidae 4863147547 50 7
Dysderidae*  4002525  0 0
Gnaphosidae* 221315034 6317
Leptonetidae* 18005050  0 0
Linyphiidae58088171003110050
Liocranidae*  3002517 13 4
Lycosidae*55810094883 63 2
Philodromidae 7450217523 8824
Salticidae* 4163146329 6320
Tetragnathidae 9463298814 7528
Theridiidae22363410019 8866
Thomisidae 753817252 7536

Discussion

the high salt marsh as an ecotone community

The high marsh can be considered an ecotone because it is a transition zone (1–80 m wide) between two adjacent ecological communities (marsh plain and upland) with a distinct plant community and unique physicochemical characteristics. More than half of the variation in community structure between marsh plain, high marsh and upland was related to a gradient in elevation, salinity, depth to anoxia and soil water content. Stressful edaphic conditions decreased with increasing elevation and the occurrence of the high marsh in the middle of the gradient supports the use of models that consider salinity and flooding when explaining salt marsh plant communities (reviewed by Ungar 1998).

Indicator species for the high marsh had maximum cover values at intermediate points on these gradients. In the more saline parts, with shallow anoxic zones and generally wetter soils, Distichlis, Jaumea, Limonium and Triglochin had higher cover than Frankenia, Grindelia, Parapholis and Polypogon. Other studies have found similar relationships between these indicator species and abiotic conditions. Distichlis spicata has been shown to have a wide breadth of tolerance to physical stress, but can be competitively excluded in more benign environments (Bertness & Ellison 1987; Zedler et al. 1992). Growth of the halophyte Frankenia salina has been shown to decrease dramatically as salinity increases (Barbour & Davis 1970). Similarly, the distribution of Parapholis (Callaway et al. 1990; Noe & Zedler 2001) and Polypogon (Kuhn & Zedler 1997; Callaway & Zedler 1998) is limited by higher soil salinity, as well as moisture stress.

Variation in substrate, freshwater inflows and land-use practices led to spatial heterogeneity between, and within, marshes for the environmental variables measured, especially for soil texture, soil wetness, soil TKN and the presence of grazing. Although the community as a whole was not significantly affected by soil texture, TKN and grazing, Frankenia and Parapholis cover was positively related to sandier soils but negatively related to soils with more silt and clay. Conversely, Jaumea was more common in soils with more silt and clay and higher water content. Randerson (1979) found that substrate texture (clay vs. sand) explained species’ distributions in a simulation model of salt marsh development. Brown (1994) found higher productivity of Jaumea along channels with silty, nitrogen-rich soils at Tomales Bay (same region as this study). Grazing by large herbivores (cattle and Tule elk) was correlated with decreased cover for all of the ecotone indicator species except the herbivore tolerant grass, Distichlis (Woolfolk 1999).

Contrary to some studies reporting an effect of spatial characteristics (e.g. birds, Helzer & Jelinski 1999), I did not find the increased plant species richness at smaller area/perimeter ratios that might be expected, particularly for exotic species that often exploit edge environments (Patton 1994). Similarly, moisture and local environmental variability, rather than habitat area, affected species richness in wetland communities of mountains in Spain (Rey Benayas et al. 1999). Ecotones between water (freshwater and marine) and land are characterized by steep gradients in physical and biological factors and effects on the characteristics of the inhabitant organisms are likely to influence biodiversity (Hansen et al. 1992). For example, plant species richness changes as salinity and soil organic matter change with salt marsh elevation (Gough et al. 1994). Elevation gradients of only a few centimetres may have profound effects on tidal inundation and thus salinity and the consequent ecotone dynamics may be more critical than biogeographical principles in describing species’ distributions in salt marshes.

the function of the high salt marsh ecotone

Ecotones may affect exchange rates of materials and organisms across community boundaries, thereby concentrating nutrients (e.g. nitrogen) and organisms (e.g. increased richness). Much management and restoration of wetland/upland borders is done on the premise that these transitions act as filters for non-point pollutants (e.g. nitrogen) although there is little evidence for changes in nitrogen pools (but see Woltemade 2000; Bai et al. 2004). The high salt marsh, with its varying redox potential and terrestrial inputs of nitrogen (e.g. nitrate run-off) has been shown to be an area that actively transforms nitrogen (Mitsch & Gosselink 1993) dependent on N-availability. Although my study did not directly measure the nitrogen cycle or retention, I did find that there was significantly more total soil nitrogen in the marsh than in the upland.

Although it is generally believed that ecotones harbour increased diversity, few empirical studies document (Hansen et al. 1992; Kark et al. 2002) or refute this (Baker et al. 2002). Both plant and spider richness increased in the high salt marsh, except at Pine Gulch, where the upper boundary was riparian forest and diversity increased with elevation. Riparian systems tend to have increased diversity with steep environmental gradients (Day et al. 1988; Pollock et al. 1998) and are often themselves characterized as ecotones. The patch dynamics between combinations of differing physical gradients (moisture, salinity) prevent a simple interpretation of ecotone dynamics at this site.

Ecotones represent the physiological limit of many species (Chabot & Mooney 1985). During much of the year, the physiological tolerances of many high salt marsh species do not permit them to thrive in either the harsh environment of the marsh plain or the nutrient-limited, competitive upland, although they can withstand the still somewhat stressful edaphic conditions of the relatively nutrient-rich ecotone. Species interactions are likely to vary as competitive abilities differ with edaphic conditions (Ungar 1998; Emery et al. 2001; Greiner La Peyre et al. 2001). Other studies have found that differences in positive and negative interactions interact with physical factors to determine salt marsh zonation (Pennings & Callaway 1992; Bertness & Shumway 1993; Bertness & Hacker 1994). Van der Maarel (1990) suggested that transition zones support species from adjacent ecosystems at the edges of their tolerance, as well as those that are at their physiological optimum within the ecotone (as seen here for Grindelia; B. H. Traut, personal observation). The ecotone between longleaf pine/wiregrass uplands and seasonal wetlands (Kirkman et al. 1998) similarly showed highest plant richness in the ecotone, whereas the upland communities had fewer species due to moisture limitations and the wetlands had fewer species due to soil saturation.

Communities at ecosystem boundaries can attract or support a variety of users, including both animal species that occupy only the ecotone and species from adjacent systems that also use the ecotone habitat for feeding or cover (Ghiselin 1977; Goeden 1979). Spider family diversity was found to be greatest in the ecotone, as found in wooded riparian zones in western Europe (Bell et al. 1999). In tidal marshes in Belgium, several spiders were found only in the marsh and therefore identified as halophilic species (Desender & Maelfait 1999). In a savanna, habitat patches had distinct communities of spider families, suggesting that patches have local specialization (Whitmore et al. 2002).

Some of the arachnids are probably using the high salt marsh for habitat and are restricted to it, whereas others are using it as a foraging ground (Ghiselin 1977; Goeden 1979), moving in and out of it throughout the day. For example, a majority of the spider families that actively hunt their prey (e.g. Anyphaenidae, Dysderidae, Gnaphosidae, Leptonetidae, Liocranidae and Salticidae), as opposed to snaring it with a web, had the highest indicator values for the ecotone, vs. the marsh plain or upland. The high salt marsh vegetation may be dense enough to provide the needed thatch for hunting. Kaston (1978) found that gnaphosids depend on dense vegetation to build their retreats. However, large hunting spiders like lycosids may prefer less dense vegetation (Döbel et al. 1990) and this may partially explain their increased abundance in the marsh plain. Because there are potentially fewer microhabitats for spiders in the less complex marsh plain, competitive interactions may also contribute to reduced diversity in the marsh plain when species with overlapping microhabitat requirements come into contact (Marshall & Rypstra 1999). In salt marsh systems, structurally complex vegetation has been shown to reduce intraguild predation and enhance co-occurrence of spiders (Finke & Denno 2002). Such an effect may have combined with increased sites for web-builders to reduce competition and enhance spider diversity.

The Araneidae, the easily recognized orb-weavers, were strong indicators of the ecotone, and probably used the high salt marsh for habitat as well as foraging grounds. Arachnid species richness and abundance, especially for web spiders, have been shown to be highly correlated with vegetation structural diversity (Greenstone 1984; Dobel et al. 1990; Gibson et al. 1992; Dennis et al. 1998; Raizer & Amaral 2001). Plants in the ecotone had the greatest variety of growth forms (graminoids, forbs, shrubs). One study that focused on wetland species found that the richness and composition of spiders was strongly influenced by the structural complexity of aquatic macrophytes (Raizer & Amaral 2001).

implications

One of the most threatened features of Pacific coast salt marshes is the high marsh/upland ecotone (James 2001). This is a unique habitat, harbouring elevated biodiversity and increased pools of total nitrogen. The terrestrial and aquatic components of biogeochemical processes, and their interaction with organisms, therefore need further study.

Ecotones have been shown on several scales to be areas of increased diversity, as well as critical habitat sites for insects (Molnár et al. 2001), arthropods (Desender & Maelfait 1999), amphibians (Dodd 1992), mammals, birds and reptiles (Zedler et al. 1992). The unique high salt marsh species found could be useful indicators for mapping and should be included in restoration. In salt marshes, ‘biocomplexity sustains biodiversity’ (Zedler et al. 2001), and spiders could be used as indicators of community structure complexity, especially as they appear to indicate the effects of land management (Gibson et al. 1992; Bell et al. 1999) and changes in plant architecture (Brandt & Lubin 1998). These results emphasize that our conservation and restoration attention and research need to focus on ecotones as well.

Acknowledgements

I am especially grateful to Dr Debbie Elliott-Fisk, Dr Michael Barbour and Dr Joy Zedler for their valuable input and guidance regarding this study, as well as critical reviews of the manuscript. Jason, Tavish and Riley Traut are thanked for their enthusiastic support, especially with fieldwork. I thank Audubon Canyon Ranch, Marin County Open Space District, Point Reyes National Seashore and Tomales Bay State Park for use of their field sites. Financial support for this study was provided by the Canon National Parks Science Scholars Program, the UC Davis Jastro Shields Grant Program, the UC Davis Consortium for Women and Research Grant and a UC Davis Graduate Research Assistantship from Dr Deborah Elliott-Fisk.

Ancillary