ecological indicators 8 (2008) 181–190
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/ecolind
Use of experimental disturbances to assess resilience along a known stress gradient Matthew G. Slocum 1, Irving A. Mendelssohn * Wetland Biogeochemistry Institute, Department of Oceanography and Coastal Sciences, Louisiana State University, Baton Rouge, LA 70803, USA
article info
abstract
Article history:
We sought to determine the effectiveness of experimental disturbances for assessing
Received 15 June 2006
resilience and stability in a Spartina alterniflora salt marsh. To do this, we applied distur-
Received in revised form
bances of different intensities along a gradient of sediment deposition that doubled as a
13 January 2007
gradient of known stress, being associated in previous studies with numerous measures of
Accepted 16 January 2007
plant vigor and soil condition. Using this gradient as a standard, we postulated a priori that areas receiving sediment were less stressed than areas which received no sediment, and therefore would be more stable and recover more rapidly after experimental disturbances.
Keywords:
For the vegetation, we found that our estimates of resilience and stability were strongly
Ecosystem health and integrity
and positively affected by sediment deposition, and therefore agreed with our a priori
Resilience
estimates. After lethal disturbance (herbicide application), vegetation in plots not receiving
Stability
sediment failed to recover, and the affected marsh turned into a mudflat and remained so
Salt and coastal marshes
during the period of observation (>2 years). In contrast, plots receiving high and moderate
Ecological indicators
amounts of sediment recovered rapidly after lethal disturbance (8–11% recovery month1 to
Sediment deposition
50% of control levels [dependent variable was a composite variable describing vegetation]). After non-lethal disturbance (trimming at the soil surface) all study plots recovered, with rate of vegetative recovery being directly associated with degree of sediment deposition. For edaphic parameters, there no was effect of disturbance, and thus these parameters appeared to be resistant to vegetation removal. These parameters appeared to be more powerfully affected by other factors, such as water level fluctuations and sediment addition. We conclude that experimental disturbances accurately assessed stress along a known stress-gradient. They also provided additional information about the underlying stress in the system. In particular, it was intriguing that in stressed areas S. alterniflora grew in elevated ‘‘hummocks’’ which appeared unable to send rhizomes into denuded areas or to recover after root death. Because recovery after experimental disturbances depends on a variety of plant-soil processes, we suggest that this recovery can be used as a bio-indicator of ecosystem condition that provides insight into important underlying determinants of structure and function. # 2007 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: +1 225 578 6425; fax: +1 225 578 6423. E-mail addresses:
[email protected] (M.G. Slocum),
[email protected] (I.A. Mendelssohn). 1 Present address: Department of Biological Sciences, Louisiana State University, Baton Rouge, LA 70803, USA. 1470-160X/$ – see front matter # 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.ecolind.2007.01.011
182 1.
ecological indicators 8 (2008) 181–190
Introduction
The world’s ecosystems are imperiled by a suite of anthropogenic stressors that jeopardize their sustainability and impair their ecological functions and societal services (Vitousek et al., 1997; Heitz, 2002). This problem has lead to an emphasis for developing accurate and rapid assessments of the underlying stress or ‘‘health’’ of ecosystems (Rapport et al., 1998). Such assessments have been increasingly defined in terms of resilience (rate of recovery after disturbance) and stability (ability of an ecosystem to maintain a steady state) (Webster et al., 1975; Leps et al., 1982). For example, several authors (Costanza, 1992; Costanza et al., 1998; Rapport et al., 1998) included resilience as one of three indices that were necessary to accurately assess underlying stress and ecosystem health (along with organization and vigor). Why are resilience and stability good indicators of underlying stress? Because they are emergent properties of ecosystems, they incorporate a suite of interacting components that provide an integrated measure of the system’s ecological status. For example, biomass recovery after a fire is dependent on the interaction of abiotic factors (e.g., fire intensity, frequency, and patch size) and biotic factors (e.g., propagule dispersal, competition, and plant life history characteristics) (Turner et al., 2003). An additional advantage of using resilience and stability as indicators of underlying stress is that they identify unknown or poorly understood stressors (Gunderson, 2000). For example, the impacts of phosphorus enrichment in the Florida Everglades did not become apparent until disturbance stimulated colonization of cattails (Davis, 1989). Although assessing resilience and stability is clearly important, it is also difficult. Some authors, for example, consider resilience to be the total ability of the system to withstand and recover from stress, and therefore to measure resilience one should analyze the system using a wide variety of stressors and disturbances (Costanza et al., 1998). However, such in depth analyzes are expensive and often not feasible, especially from a management standpoint; there are just too many valuable and imperiled ecosystems whose stress needs to be estimated. Perhaps one solution to this problem is to assess resilience using experimental disturbances, an approach that has been used by a number of authors in various ecosystems (Cole, 1995; Matthaei et al., 1996; Walker et al., 1997; Lavorel, 1999; Cobb et al., 2001). However, to our knowledge no studies have tested how effectively experimental disturbances estimate underlying stress. This requires that the estimates of underlying stress derived from experimental disturbances be compared to other stress estimates (see Whitford et al., 1999, who used a natural disturbance to assess resilience along a known stress gradient). If there is agreement between a priori stress estimates and the results of the disturbances, then the disturbances can be used to access underlying stress in this ecosystem type. The experimental disturbances will have a better chance of revealing information about resilience if they mimic characteristics (e.g., intensity, scale) of natural and anthropogenic disturbances that commonly occur in the ecosystem. Also, because some components of the ecosystem may be more resilient than others (Westman, 1978), multiple
indicators and disturbances estimates should be used to provide more information about the system’s underlying stress. In this study, we applied disturbances along a stress gradient that has been studied since the early 1990s (Kuhn and Mendelssohn, 1999; Mendelssohn and Kuhn, 1999, 2003; Slocum et al., 2005). In order to gain the maximum information about the resilience and stability along this gradient, two intensities of disturbance were applied, and numerous potential indicators of resilience and stability monitored. We postulated that more stressed areas would recover more slowly than less stressed areas, and that failure to recover would be more likely in more stressed areas, especially after more intense disturbance. Our goal was to outline a method for assessing resilience and stability that is rapid, accurate, widely applicable, and that provides a measure of underlying stress and ecosystem health. This measure of resilience and stability could then be included with other indicators (e.g., measures of vigor and organization; Costanza et al., 1998) that together represent a comprehensive measure of the health of an ecosystem.
2.
Methods
2.1.
Study site
Our study site (29811.850 N, 89826.300 W) was a degrading, microtidal salt marsh located 13 km southwest of the town of Venice, Louisiana. This marsh lies within the Modern (Birdsfoot) Delta of the Mississippi River Delta Complex and has rates of relative sea level rise nine times that of worldwide eustatic rise (0.94 cm year1 at Port Eads, Louisiana, from 1944 to 1988; Penland and Ramsey, 1990). These rates have resulted in land loss rates of 5.7–11.4 km2 year1, some of the highest in the delta (Dunbar et al., 1992). In January 1992, there was a mitigation project in the area to fill a canal using sediment hydraulically dredged from the Gulf of Mexico. Some sediment accidentally overflowed into a deteriorating salt marsh dominated (>99% cover) by Spartina alterniflora Loisel. This spillage created a sediment-deposition gradient that reached 40 cm above the original marsh surface. (Increase in elevation was measured relative to nearby marsh that did not receive sediment, and is hereafter referred to as relative elevation.) With time, the sediment consolidated and compacted such that by 1998 relative elevations ranged from 0 to 22 cm.
2.2.
Experimental design and hypotheses
The gradient was related to several indicators of stress known to be important in salt marshes, including plant vigor, soil redox potential, bulk density, sulfide concentrations, and duration and depth of flooding (Kuhn and Mendelssohn, 1999; Mendelssohn and Kuhn, 1999, 2003; Slocum et al., 2005). Based on these studies of plant vigor and soil physico-chemical condition, we divided the gradient into three zones that varied in these stressors, including a no deposition zone, a moderate deposition zone, and a high deposition zone (Table 1). The no deposition zone was clearly the most highly stressed, having the least biomass and cover, low soil bulk density, highly
183
ecological indicators 8 (2008) 181–190
Table 1 – Measures of plant vigor and soil condition in the three deposition zones from previous studies* Vigor indicator
Vegetation cover (%) Total biomass (g 0.25 m2) Stand height (cm) Soil salinity (ppt) Sulfide (mM) Bulk density (g cm3) Eh at 2 cm depth (mV) H2O depth (cm) Elevation (cm) Sand (%) *
Data years
97, 97, 97, 97, 97, 97, 97, 97, 98 96
98 98 98 98 98 98 98 98
Deposition zone No
Moderate
37 14 259 88 97 6 13 1.7 1.1 0.6 0.27 0.06 147 38 10.6 16.1 2.4 2.5 78
72 14 272 101 118 23 19 1.9 0.02 0.02 0.53 0.15 134 117 1.1 4.1 3.3 0.7 22 14
High 61 13 255 29 98 12 18 1.5 0.03 0.003 0.86 0.21 13 147 6.1 5.9 13.4 4.1 58 13
Mendelssohn and Kuhn (1999, 2003); Kuhn and Mendelssohn (1999); Slocum et al. (2005).
reduced soils, high concentrations of hydrogen sulfide, and high flooding depth. The moderate deposition zone was the least stressed, having low concentrations of hydrogen sulfide, intermediate levels of sand, good bulk density, and high percent cover and high total biomass. The high deposition zone had many of the characteristics of the moderate deposition zone, but it had more sand and slightly less cover, stand height, and live biomass estimates than the moderate deposition zone. Therefore a priori we ranked the zones as: moderate deposition zone > high deposition zone low deposition zone. For more information on how these zones are ranked, see Slocum et al. (2005). Within each zone, we assessed how indicators of stability and resilience responded to two intensities of disturbance to S. alterniflora stands. Disturbance to S. alterniflora is clearly relevant, as it is the ‘‘foundation species’’ (Dayton, 1972) that biologically generates and maintains the habitat (Bruno and Kennedy, 2000). The disturbances were similar to those of natural disturbances that the system normally experiences, and included: (1) Non-lethal disturbance—above-ground vegetation was clipped to the soil surface using a gasolinepowered hand-held trimmer on 5/9/2000 and again on 6/30/ 2000. This disturbance mimicked herbivory. (2) Lethal disturbance—vegetation was sprayed with Roundup herbicide (the isopropylamine salt of glyphosate; Monsanto, St. Louis, Missouri, USA) in a water/detergent solution at recommended levels on the same dates as the non-lethal applications. After the herbicide applications, all vegetation within the plots was browned. This disturbance mimicked situations in which there is total plant mortality, such as when wrack is deposited and then subsequently removed by tides (Brewer et al., 1998). The disturbances were applied to each zone in five randomly located blocks, each consisting of three plots, one for each disturbance intensity and one as a control. The plots were 2 m 2 m in size and separated by a 1 m buffer. This plot size was large enough such that, based on known recovery rates of S. alterniflora in healthy marshes (I.A. Mendelssohn, unpublished data), the plots would easily fill in within two years unless the area was significantly stressed. The plots were not so small, however, as to fill in too rapidly such that resilience and stability would not be discernable. The experiment therefore had 45 plots (3 zones 5 blocks 3 disturbances). These plots were sampled three times (9/28/
2000, 5/9/2001, and 10/30/2001) to quantify rate of recovery. In this time series, we considered the last day of disturbance (6/ 30/2000) as time zero, and the different sampling periods as 3, 7, and 13 ‘‘growth’’ months, respectively, with a ‘‘growth’’ month not including winter months (December–February). We base growth months on our observations during numerous studies that S. alterniflora in the region grows little in December through February (Kuhn and Mendelssohn, 1999; Mendelssohn and Kuhn, 1999, 2003; Slocum et al., 2005). For vegetation, we hypothesized: (1) highly stressed plots would be unstable (would not recover after disturbance), especially after more intense disturbance. (2) Given stable plots, highly stressed plots would have lower resilience (slower recovery rates) than less stressed plots. For the soils we hypothesized: (1) some soil traits might be resistant to the effects of vegetation removal, particularly if there were other factors that were more important in governing these traits (e.g., periodic flooding). (2) If the soils were not resistant, then vegetation removal could decrease soil redox potential (less roots to transport oxygen), increase nutrients (less uptake by plants), increase salinity (increased exposure to sun and wind), and increase decomposition rates (because of increased nutrient concentrations). (3) If the vegetation did not recover, these edaphic trends could become reinforced with time until they reached some endpoint different from control levels. The degree of change from the previous steady state could then be measured. (4) In contrast, if the vegetation recovered, these trends would likely reverse, and therefore resilience could be measured. Note that for our study, detecting stability may not be possible if recovery of some plots was slow and incomplete by 13 ‘‘growth’’ months. This is a limitation to all studies of stability, because it is difficult to determine when recovery is complete and an experiment is actually over (Underwood, 1989, p. 68). However, the main goal of this study was not to provide absolute estimates of stability or resilience, but rather to use relative differences in stability and resilience to assess underlying stress. For this purpose, we found that 13 ‘‘growth’’ months was adequate to provide relevant information. Similarly, estimating resistance of soils is difficult because there may be subtle but statistically undetected shifts in soil condition in response to disturbance. However, again, the purpose of this study was not to comprehensively evaluate resistance. Rather, we seek to determine if there are large and
184
ecological indicators 8 (2008) 181–190
important shifts in soil condition in response to disturbance, and if these shifts vary based on the stress gradient. Because the goal of this research was to test the method of experimental disturbances as a means of assessing habitat condition (stability and resilience), the evaluation of the sediment-overflow area was used as a case study. Hence, statistical inference based on this study site, which could not be replicated in another marsh, applies to this particular study area. However, if the assessment method succeeded in this salt marsh, we have no reason to believe it would not be successful in other salt marshes and herbaceous habitats.
2.2.1.
Sampling and indicators
Two sets of indicators – vegetation and soils – were examined during each sampling period. Although these indicators by themselves cannot provide a comprehensive measure of resilience and stability, they are two critical components of the ecosystem (see Mitchell et al., 2000). To examine these indicators, each 2 m 2 m plot was divided into 16 sub-plots (0.25 m2). Of these sub-plots, three were randomly chosen for sampling, one for each sampling period. Within each sub-plot, the following were measured: (A) Vegetation. All above-ground tissue was clipped at the ground surface. Live material was examined to determine number of stems and mean canopy height. It was then dried to constant weight at 65 8C, and weighed to determine biomass. (B) Soil decomposition. Decomposition in soils was evaluated using the cotton strip assay (Harrison et al., 1988). In each sub-plot, a 10 cm 30 cm strip of standardized cotton cloth (Shirley Dyeing and Finishing Ltd., UK) was inserted vertically and length-wise into the soil with a spade, and the level of the soil surface was marked with a cut. The strips were retrieved after 14 days, and on the same day reference cotton strips were inserted into the soil and immediately removed. All strips were then washed in deionized water until clean, dried at 25 8C, and reduced to 2 cm wide sub-strips by cutting and fraying. The sub-strips were then moistened and pulled apart at 25 8C to test their tensile strength using a Dillon Quantrol Snapshot tensometer and a force gauge. Each sub-strip’s tensile strength was calculated relative to the mean of the sub-strips of the reference strips. From these calculations, we derived cotton tensile strength loss (CTSL) on a percent loss per day basis. For each strip, the CTSL values for its sub-strips were averaged for analysis. (C) Soil redox potential (Eh). Eh was determined at 2 and 15 cm soil depth with three bright platinum electrodes and a calomel reference electrode. Three sub-samples were taken at each depth, and their mean was used for statistical analysis. Measured potential was normalized to the standard hydrogen electrode by adding 244 mV. (D) Exchangeable ions. A core (5 cm diameter by 12 cm deep) was collected, immediately sealed in a water-tight plastic bag, homogenized, and stored at 4 8C. Exchangeable P was extracted by mixing 2 g of soil with 40 ml of 0.03 M NH4–F and 0.1 M HCl (Byrnside and Sturgis, 1958). Exchangeable Na, K, Mg, and Ca were extracted by mixing 2 g of soil with 20 ml of 1 M NH3OAc at neutral pH (Thomas, 1982).
Exchangeable Mg, Fe, and Zn were extracted by mixing 10 g of soil with 20 ml 0.005 M DTPA at 7.3 pH (Baker and Amacher, 1982). Elemental concentrations in the extractions were determined on a Spectro CirosCCD inductively coupled argon plasma emission spectrophotometer (ICP). (E) Interstitial water characteristics. A core (5 cm diameter by 12 cm tall) was collected, and immediately sealed in an airtight centrifuge bottle and purged with nitrogen. Each bottle was stored at 4 8C for no more than 24 h. It was then centrifuged and its water removed. This water was used to measure: (1) Salinity using a salt refractometer with automatic temperature compensation. (2) pH with a gelfilled combination electrode. (3) Elemental concentrations (Al, Ca, Fe, K, Mg, Mn, Na, P, and S). This sample was filtered (0.45 mm), preserved with a drop of 50% concentrated nitric acid, and stored at 4 8C until analyzed. Elemental concentrations were determined on the ICP. (4) Sulfide: this sample was mixed with antioxidant buffer and analyzed with a sulfide electrode (Lazar Research Laboratories, Los Angeles, California). (5) NH4–N: this sample was filtered (0.45 mm) and frozen until analyzed on a Lachat Quickchem 8000. (F) Bulk density and percent soil moisture. A core (5 cm diameter by 10 cm tall) was collected and dried to a constant weight at 70 8C. Derived data was used to convert soil nutrient concentrations to a volumetric basis, which is preferable when there is a high variation in bulk density among sampling sites (DeLaune et al., 1979).
2.3.
Data analysis
Because of numerous dependent variables, they were first decomposed into composite variables using principle components analyses (PCAs). Separate PCAs were conducted for the vegetation and soil variables. Factors scores from components that explained over 10% of the data set’s variation were used in subsequent analyses. For the vegetation, plots during the last sampling period were first tested for stability. To determine this, the vegetation in each disturbed plot was compared to the 95% confidence interval around the mean of the control plots of that deposition zone. If the vegetation in a plot did not reach within this confidence interval, it was considered to be unstable. This procedure was done for the scores of each ‘‘vegetation’’ principal component. For each deposition zone, we then determined if the number of plots not recovering differed from what would be expected given stability (i.e., all plots recovering) using a x2 test. In each stable plot, resilience (recovery rate) was determined using a quadratic regression. The dependent variable was percent of control, which was the plot’s factor score divided by the mean factor score of the controls for its deposition zone during the same sampling period. The independent variables were time and time squared, and the y-intercept was set to zero (there was no vegetation at the beginning of the experiment). Factor scores were set to have a minimum value of zero. Once a regression equation for a plot was derived, we used it to estimate the rate at which the plot’s factor score reached 50% of control values (i.e., % recovery month1 to 50% of controls). This ‘‘rate to 50%’’ approach was used (rather than estimating the time it took for
185
ecological indicators 8 (2008) 181–190
full recovery) because recovery in some plots approached 100% without fully reaching 100%. Once recovery rates for the stable plots were derived, we determined how deposition zone and disturbance affected recovery rate. This determination was done separately for each disturbance treatment using a one-way mixed-model ANOVA, with deposition zone as the main effect and block (nested within zone) as a random effect. We did not conduct a two-way ANOVA (including both zone and disturbance treatment) simply because there was a sufficient lack of recovery in particular zone/disturbance treatment combinations to cause problems with this approach (see Section 3). Because there were two ANOVAs, type one error was corrected to 0.025 (Bonferroni adjustment). For soils, we first checked for resistance by examining each ‘‘soils’’ principal component with a repeated-measures mixed-model factorial. This factorial included the repeated effect of sampling period, the fixed effects of deposition zone and disturbance, block (nested within zone) as a random effect, and the appropriate interactions. If disturbance had no significant effect in at least one of the deposition zones or sampling periods, we assumed that the soils were resistant to disturbance. If the scores of a component were not resistant, we examined for stability and resilience in the same way we did for the vegetation. Note that, in these repeated-measures factorial analysis of soils, deposition zone and sampling period are included as main effects solely for the purpose of removing variation associated with them. For this study, we are interested in differences among the disturbance treatments and whether the effects of these disturbances vary by deposition site. Main effects of deposition sites are not of interest in this study since they have already been thoroughly described (Kuhn and Mendelssohn, 1999; Mendelssohn and Kuhn, 1999, 2003; Slocum et al., 2005).
For all statistical tests, normality and homogeneity of variance were judged by examining residuals using Shapiro– Wilk statistics and box-plots, and natural log or rank transformations were used when necessary to improve normality and homogeneity of variance (Conover and Iman, 1981). Pairwise comparisons were made using Tukey– Kramer HSD tests. All statistics were performed using the MIXED, FACTOR, and REG procedures of SAS 6.12 (SAS Institute, 1990).
3.
Results
3.1.
Stability of the vegetation
After lethal disturbance, recovery of above-ground live biomass, number of stems, and mean canopy height of S. alterniflora in the high and moderate deposition zones surpassed control levels by 12–45% (Table 2). In contrast, the no deposition zone had no recovery. To test the significance of these results, we used a component (derived from PCA) which described 84% of the variation in the data set and on which all of the measures of vegetation were highly loaded (Table 3A). Using the factors scores from this component, we found that, after lethal disturbance, all of the plots of the high and moderate deposition zones had fully recovered compared to control levels (i.e., they had reached within the 95% confidence limit of the controls after 13 ‘‘growth’’ months) (Table 2). For the no deposition zone, however, none of the plots had reached within the 95% confidence interval of the control plots (a significant difference; x21 ¼ 5, P = 0.025; Table 2). After non-lethal disturbance, all three deposition zones were stable, as all plots in the study recovered to within the 95% confidence interval of the control plots by the last sampling period (Table 2).
Table 2 – Means W 1 S.E. of variables describing recovery of Spartina alterniflora in 0.25 m2 plots 13 ‘‘growth’’ months after application of disturbances in three zones differing in degree of sediment deposition. Within a row, mean percent of control values 1 S.E. for the non-lethal and lethal treatments are included in parentheses Deposition zone
Disturbance treatment Control
Non-lethal
No deposition Above-ground biomass (dry g) Number of stems Mean canopy hgt (cm) Factor score
292 37 109 12 104 5 1.1 0.2
272 20 (93 7%) 98 20 (90 18%) 107 6 (103 6%) 0.9 0.2 (94 7%)NS
Moderate deposition Above-ground biomass (dry g) Number of stems Mean canopy hgt (cm) Factor score
261 22 73 14 97 4 0.7 0.2
255 35 (98 13%) 73 3 (100 4%) 93 5 (96 5%) 0.6 0.2 (95 7%)NS
380 56 (145 21%) 101 10 (138 14%) 109 9 (112 9%) 1.3 0.3 (124 11%)NS
High deposition Above-ground biomass (dry g) Number of stems Mean canopy hgt (cm) Factor score
219 27 78 10 77 8 0.3 0.2
197 12 (90 5%) 78 8 (100 10%) 78 8 (101 10%) 0.3 0.1 (97 2%)NS
286 35 (130 16%) 97 9 (124 11%) 97 7 (125 9%) 0.8 0.2 (121 9%)NS
NS *
Lethal 0 0 (0%) 0 0 (0%) 0 0 (0%) 1.9 0.0 (0%)*
Not significalty different from controls within the deposition zone (all plots grew to within the 95% confidence interval of the controls). Significantly different from controls within the deposition zone ðX21 ¼ 5; P ¼ 0:025Þ.
186
ecological indicators 8 (2008) 181–190
Table 3 – Correlations between indicator variables and principal components (PCs), including one principal component for the vegetation (A) and three for the soils (B) Indicator variables
(A) Vegetation PC 1
(B) Soils PC 1: reduction
PC 2: salinity
Exchangeables Calciumln Phosphorus Potassium Zinc Sodium Manganeseln Iron Magnesium
0.65 0.64 0.47 0.37 0.31 0.25 0.11 0.05
0.07 0.21 0.28 0.30 0.01 0.09 0.56 0.52
0.40 0.42 0.03 0.45 0.08 0.06 0.05 0.01
Interstitial water Potassium Ironln Magnesium Calcium Sodium Sulfurln Aluminum Ammoniumln Manganese Phosphorus
0.79 0.69 0.12 0.21 0.51 0.10 0.00 0.48 0.40 0.27
0.18 0.23 0.94 0.88 0.62 0.12 0.05 0.51 0.04 0.07
0.06 0.33 0.03 0.21 0.13 0.86 0.64 0.41 0.01 0.50
Vegetation Above-ground biomass Number of stems Mean canopy height
PC 3: decomposition
0.91 0.92 0.91
Miscellaneous Eh (2 cm depth) Total dissolvable sulfides Eh (15 cm depth) Bulk density % Soil moisture Salinity Decomposition rate pH
0.80 0.77 0.73 0.69 0.63 0.17 0.11 0.05
0.31 0.33 0.34 0.45 0.56 0.93 0.20 0.34
0.06 0.13 0.03 0.25 0.27 0.04 0.85 0.13
Eigenvalue % Variance explained
2.5 84
9.6 35
4.1 15
2.6 10
Also included are eigenvalues and percent variation explained. Bolded scores indicate variables that define the component. Natural log transformed before analysis.
ln
Therefore, our tests suggest that disturbed plots of the high and moderate deposition zones were stable, regardless of disturbance intensity. In contrast, plots in the no deposition zone were only stable after low intensity disturbance, and were unstable after a higher intensity disturbance that caused plant mortality.
3.2.
Resilience of the vegetation
After lethal disturbance, the high deposition zone had faster recovery than the moderate zone (one-way ANOVA on natural log transformation values: df = 1.8; F = 7.9; P = 0.02; significant after Bonferroni adjustment for two tests) (Fig. 1A). (Resilience after lethal disturbance in the no deposition zone was not included in this test because these plots did not recover, i.e., they were not stable.) After non-lethal disturbance, the deposition zones also differed in recovery rate (one-way ANOVA after rank transform; df = 2.15; F = 5.1; P = 0.02; statistically significant after
Bonferroni adjustment for two tests) (Fig. 1B). Recovery in the no deposition zone was significantly less rapid than that in the high zone (Tukey–Kramer HSD, P < 0.05). The moderate deposition zone’s recovery rate was intermediate and not statistically different from those of the high and no deposition zones (Tukey–Kramer HSD, P > 0.05). These tests on the resilience of the vegetation therefore reinforced the results of the stability trials, showing that the no deposition zone was more highly stressed. Contrary to our predictions, however, it appeared that the high deposition zone had higher resilience than the moderate deposition zone, as it recovered faster after lethal disturbance.
3.3.
Resistance of the soils
We tested the effect of disturbance on factor scores of principal components that described soils (Table 3B). These components included one that described soil reduction and sulfide concentrations (and which explained 35% of the
ecological indicators 8 (2008) 181–190
Fig. 1 – Recovery rates of vegetation among three sediment deposition zones (no, moderate, and high deposition) after (A) lethal disturbance and (B) non-lethal disturbance. Bars that do not share the same letter are significantly different at P = 0.05 (Tukey-Kramer HSD tests).
variation in the data set), a second component that described salinity (15%), and a third component that described decomposition (10%) (Table 3B). In addition, we examined interstitial NH4–N separately because it is usually a limiting nutrient in these ecosystems (Mendelssohn and Morris, 2000), and because it was not well represented on any of the components. The tests on the components and on interstitial NH4–N revealed that the soil parameters were not substantially affected by disturbance (Table 4). It therefore appeared that vegetation removal had little or no effect (i.e., not detected statistically) for these soil parameters, and that they were largely resistant to disturbance at this study site (Table 5).
4.
Discussion
4.1.
Stability and resilience over the stress gradient
We applied disturbances along a stress gradient to determine how the resultant estimates of resilience corresponded with a priori estimates of stress based on previous studies (Mendelssohn and Kuhn, 1999, 2003; Kuhn and Mendelssohn, 1999;
187
Slocum et al., 2005). In this way, we hoped to test if the experimental disturbances were accurately describing underlying stress. In our first a priori prediction (high and moderate deposition zones no deposition zone), we found that the applied disturbances resulted in estimates of resilience and stability that agreed with the prediction. In response to nonlethal disturbance, the no deposition zone had lower resilience (slower recovery rates) than the high zone, and after lethal disturbance the no deposition zone did not recover during the study period, and remained denuded. Therefore, the disturbances successfully measured resilience and stability, indicating that resilience and stability can describe underlying stress when distinguishing between sites of such disparity. Our second a priori prediction (moderate zone > high zone) was not supported by the data. Our prediction of the resilience of the high deposition zone was based on it having somewhat more sand and somewhat less plant vigor than the moderate deposition zone (Table 1). The higher resilience of the high deposition zone, however, may have been because of its being in transition from a high to low vigor state. One year after sediment addition (in 1992), the high deposition zone had much greater biomass and cover than the moderate deposition zone (Mendelssohn and Kuhn, 2003). Over time, however, the vigor of S. alterniflora in the high zone decreased, so that by 1998 its plant vigor was somewhat less than that in the moderate zone (Table 1; Slocum et al., 2005). It is likely that the somewhat greater resilience of the high deposition zone was a reflection of this initially greater vigor. The rapid recovery of S. alterniflora after non-lethal disturbance demonstrates its ability to store carbohydrate and nutrient reserves below-ground and to thereby rapidly produce tillers after winter senescence (Hopkinson and Schubauer, 1984; Gallagher and Howarth, 1987) or after herbivory and burial under wrack or sediment. S. alterniflora’s ability to store reserves underground also allows it to rapidly and vegetatively colonize areas denuded by more intense disturbance (e.g., deeper burial), and also after our lethal disturbances in the high and moderate deposition zones. The response of the vegetation after lethal disturbance in the high and moderate deposition zones was particularly impressive in that maximum recovery substantially surpassed that of control levels (Table 2). This was likely a response to a pulse of released nutrients soon after plant mortality. This response has been documented previously in salt marshes recovering from natural dieback (McKee et al., 2004). We did not likely detect this pulse in plant-available nutrients because of the rapid response of the vegetation for uptaking released nutrients, especially ammonium. Unlike the high and moderate deposition zones, the no deposition zone did not recover after lethal disturbance, possibly because neighboring plants were impeded from sending rhizomes through soil that had sulfide concentrations (1.6 0.1 mM) above that known to stress this species (1.0 mM) (Koch and Mendelssohn, 1989; Bradley and Morris, 1990; Koch et al., 1990). We observed that in the no deposition zone the plants grew in ‘‘hummocks’’ constructed of roots, rhizomes, and detritus, and that between the hummocks there were no rhizomes or roots. These hummocks may be an adaptation for survival in waterlogged conditions by maintaining plants at a higher elevation than the surrounding soil. If so, it may explain
188
ecological indicators 8 (2008) 181–190
Table 4 – Means W 1 S.E. for several variables describing the soils of the deposition zones, disturbance treatments, and sampling periods (period 1 = 9/28/2000, period 2 = 5/9/2001, period 3 = 10/30/2001) No deposition Control
Non-lethal
(A) Soil reduction Sulfide (mM) Period 1 1.5 0.3 Period 2 1.1 0.4 Period 3 2.2 0.4 Eh (mV at 2 cm soil depth) Period 1 123 17 Period 2 32 33 Period 3 63 16
Moderate deposition Lethal
Control
Non-lethal
Lethal
Control
Non-lethal
Lethal
1.9 0.1 0.8 0.3 2.1 0.1
1.8 0.2 0.8 0.2 2.3 0.4
0.0 0.0 0.0 0.0 0.2 0.1
0.0 0.0 0.0 0.0 0.1 0.0
0.0 0.0 0.0 0.0 0.2 0.0
0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0
0.1 0.0 0.0 0.0 0.0 0.0
116 7 59 19 93 3
107 13 63 12 95 20
58 20 5 36 29 9
32 36 46 22 5 34
44 23 45 21 37 16
82 13 171 9 201 40
155 21 196 46 148 21
177 43 164 28 172 24
17 0 14 1 23 1
17 1 14 1 24 0
14 1 15 1 25 1
15 2 23 1 22 1
21 1 24 2 22 1
18 1 24 1 25 1
4.0 0.1 4.9 0.1 5.0 0.2
4.0 0.2 4.9 0.1 4.6 0.4
4.3 0.1 5.0 0.2 4.9 0.1
3.9 0.2 6.3 0.1 4.9 0.2
4.1 0.1 6.2 0.1 5.1 0.2
4.2 0.2 6.5 0.1 4.8 0.3
11 1 33 18 35 10
18 6 31 12 27 6
81 48 25 26 4
53 26 13 4 36 17
57 28 18 4 19 5
38 25 18 6 21 3
(B) Salinity: interstitial salinity (ppt) Period 1 12 0 12 1 Period 2 10 0 10 0 Period 3 16 1 16 1 (C) Decomposition: cotton tensile Period 1 4.6 0.0 Period 2 5.6 0.2 Period 3 5.6 0.3
12 1 90 15 1
strength loss (% day1) 4.6 0.0 4.7 0.0 6.0 0.2 6.3 0.0 5.8 0.2 6.2 0.1
(D) Nitrogen: interstitial NH4–N (mM) Period 1 126 4 126 3 Period 2 556 240 796 297 Period 3 326 122 186 111
124 3 845 376 654 41
why S. alterniflora could not tolerate the death of its roots in this zone. Indeed, more than two years after disturbance was applied, we observed that the plots still had not recovered. It therefore appears that these sites had changed to mudflats, at least during the period of observation. In contrast to the vegetation, we did not detect a response of soil parameters to disturbance. This lack of a response may have been due to environmental factors, such as water level fluctuations and sediment quality, which may have overwhelmed more subtle effects caused by vegetation removal.
4.2.
High deposition
Disturbance intensity and multiple indicators
Because some components of an ecosystem may be more resilient than others (Westman, 1978), we sought to determine if multiple indicators and disturbance intensities provided
more information about resilience and stability along the stress gradient than just a few indicators or just one disturbance intensity. Two intensities of disturbances resulted in a more sensitive assessment of resilience and stability than would have occurred if we had just used one. It showed that in the no deposition zone, S. alterniflora was capable of recovering after non-lethal disturbance but not after lethal disturbance. This result suggests that S. alterniflora was not capable of recovering under highly stressed conditions, and it further suggests the importance of hummocks as an adaptation for surviving these conditions. While two disturbance intensities were informative, in this study the use of multiple indicators was not. For the vegetation, measuring percent cover and height of the system’s ‘‘foundation species’’ S. alterniflora (Dayton, 1972; Bruno and Kennedy, 2000) would have been sufficient to
Table 5 – Repeated-measures mixed-model factorials of three principal components (PCs) describing soils in a salt marsh near Venice, LA, USA Source
Deposition zone Sampling period Zone period Disturbance Zone disturbance Period disturbance Zone period disturbance
df
2.12 2.56 4.56 2.56 4.56 4.56 7.56
PC 1: reduction
PC 2: salt
PC 3: decomposition
Interstitial NH4–N
F
P
F
P
F
P
F
142 4.0 32 1.3 0.7 1.7 0.7
0.0001 0.02 0.0001 0.29 0.49 0.15 0.67
149 111 12 0.5 0.2 0.6 1.6
0.0001 0.0001 0.0001 0.63 0.95 0.67 0.16
19 167 25 1.1 0.3 0.5 0.6
0.0002 0.0001 0.0001 0.32 0.84 0.77 0.73
39 1.3 6.1 0.3 0.7 1.1 0.8
P 0.0001 0.28 0.0002 0.76 0.59 0.34 0.61
Also included is an analysis of interstitial ammonium. Effects in the models included those of interest in the study (disturbance and interactions with disturbance) and those not of interest (deposition zone, sampling period, and their interaction).
ecological indicators 8 (2008) 181–190
assess resilience. As for soil characteristics, it did not appear that applying disturbances was an efficient tool for studying the site’s underlying stress. Vegetation removal certainly affects soil characteristics in salt marshes, but at our microtidal site it appeared that the soils were highly influenced by other factors such as wind-induced inundation, and that to decouple the effects of vegetation removal and these other factors, considerable statistical power would have been required.
4.3. Conclusions: use of experimental disturbances to assess resilience As opposed to opportunistic studies of anthropogenic and natural disturbances, experimental disturbances allow researchers to plan the timing, scale, frequency, intensity, and experimental design of studies of resilience and stability. For this reason they have been used in numerous studies (e.g., Cole, 1995; Matthaei et al., 1996; Walker et al., 1997; Lavorel, 1999; Cobb et al., 2001; Patrı´cio et al., 2006). Another advantage of experimental disturbances is that they can assess the impacts of natural and anthropogenic disturbance before they happen (Underwood, 1989). They thus can be used to assess the stress of ecosystems with valuable species or services that are threatened by potential catastrophes such as hurricanes or oil spills. Experimental disturbances would also be useful for accessing the stability of newly restored ecosystems. For example, in Louisiana the ability of restored marshes (such as the one in this study) to survive hurricanes is not known. The use of disturbances such as the ones we applied provides more information on how well restored marshes will respond to such a disturbance. In short, we believe that for salt marshes, our experimental disturbances, or similar ones, can be used by researchers and managers to test underlying stress. Based on the data from these experiments, managers can decide when and where to restore. Similarly, experimental disturbances can be used successfully in other ecosystems. This may initially require examination of a broader range of indicators (e.g., recovery of species composition), or a longer period of study to address longer recovery periods (e.g., forests). Although experimental disturbances as resilience ‘‘probes’’ provide important information, they should not be used as the sole evaluation of stress or ecosystem health because ecosystem health is also dependent on other processes. Rather, they should be used in concert with other indicators, such as measures of system organization (e.g., species diversity and trophic interactions) and vigor (e.g., soil and plant metabolism) (Costanza, 1992; Costanza et al., 1998; Rapport et al., 1998). Furthermore, use of experimental disturbances as resilience probes should address the effect of disturbance intensity, severity, and scale on a variety of structural attributes, functional processes, and trophic levels (Jørgensen, 2002; Patrı´cio et al., 2006).
Acknowledgements For their primary financial support, we would like to thank the Louisiana Sea Grant College Program, a part of the National
189
Sea Grant College Program, and the Coastal Ocean Program, both maintained by the National Oceanographic and Atmospheric Administration, United States Department of Commerce. We also thank Dr. James Pahl, Chris Anastasio, Dr. Lee Stanton, Pamela Weisenhorn, and Dr. Megan LaPeyre for fieldwork. Dr. Jay Geaghan (Department of Experimental Statistics, Louisiana State University) helped with data analysis, and Tom Oswald (Coastal Ecology Institute, Louisiana State University) and Mike Breithaupt (Soil Testing and Plant Analysis Laboratory, Louisiana State University) helped with soils analyses. T. Baker Smith & Sons conducted elevation surveys. Lastly, we thank the anonymous reviewers for their comments.
references
Baker, D.E., Amacher, M.C., 1982. Nickel, copper, zinc, and cadmium. In: Page, A.L., et al. (Eds.), Methods of Soil Analysis. American Society of Agronomy, Madison, Wisconsin, USA, pp. 323–336. Bradley, P.M., Morris, J.T., 1990. Influence of oxygen and sulfide concentration on nitrogen uptake kinetics in Spartina alterniflora. Ecology 71, 282–287. Brewer, J.S., Levine, J.M., Bertness, M.D., 1998. Interactive effects of elevation and burial with wrack on plant community structure in some Rhode Island salt marshes. Journal of Ecology 86, 125–136. Bruno, J.F., Kennedy, C.W., 2000. Patch-size dependent habitat modification and facilitation on New England cobble beaches by Spartina alterniflora. Oecologia 122, 98–108. Byrnside, D.S. Jr., Sturgis, M.B., 1958. Soil phosphorus and its fractions as related to response of sugar cane to fertilizer phosphorus. Bull No. 513, Louisiana State University Agriculture Experiment Station, Baton Rouge, Louisiana, USA. Cobb, A.R., Nadkarni, N.M., Ramsey, G.A., Svoboda, A.J., 2001. Recolonization of bigleaf maple branches by epiphytic bryophytes following experimental disturbance. Canadian Journal of Botany 79, 1–8. Cole, D.N., 1995. Experimental trampling of vegetation II. Predictors of resistance and resilience. Journal of Applied Ecology 32, 215–224. Conover, W.J., Iman, R.I., 1981. Rank transform as a bridge between parametric and nonparametric statistics. The American Statistician 35, 124–133. Costanza, R., 1992. Toward an operational definition of health. In: Costanza, R., Norton, B., Haskell, B. (Eds.), Ecosystem Health, New Goals for Environmental Management. Island Press, Washington, DC, USA, pp. 239–256. Costanza, R., Mageau, M., Norton, B., Patten, B.C., 1998. Predictors of ecosystem health. In: Rapport, D., Costanza, R., Epstein, P.R., Gaudet, C., Levins, R. (Eds.), Ecosystem Health. Blackwell Science, Inc., Malden, MA, USA, pp. 240–250. Davis, S.M., 1989. Sawgrass and cattail production in relation to nutrient supply in the Everglades. In: Fresh Water Wetlands and Wildlife, 9th Annual Symposium, Savannah River Ecology Laboratory (United States Department of Energy), Charleston, SC, USA, pp. 24–27. Dayton, P.K., 1972. Toward an understanding of community resilience and the potential effects of enrichment to the benthos at McMurdo Sound, Antarctica. In: Parker, B.C. (Ed.), Proceedings of the colloquium on conservation problems in Antarctica, Allen Press, Lawrence, KS, USA, pp. 81–95.
190
ecological indicators 8 (2008) 181–190
DeLaune, R.D., Buresh, R.J., Patrick Jr., W.H., 1979. Relationship of soil properties to standing crop biomass of Spartina alterniflora in a Louisiana marsh. Estuarine and Coastal Marine Science 8, 477–487. Dunbar, J.B., Britsch, L.D., Kemp III., E.B., 1992. Land loss rates. Report 3: Louisiana coastal plain. U.S. Army Corps of Engineers, Waterways Experiment Station, Technical Report GL-90-2, Vicksburg, Mississippi. Gallagher, J.L., Howarth, R.W., 1987. Seasonal differences in Spartina recoverable underground reserves in the Great Sippewissett Marsh in Massachusetts. Estuarine Coastal and Shelf Science 25, 313–319. Gunderson, L.H., 2000. Ecological resilience – in theory and application. Annu. Rev. Ecol. Syst. 31, 425–439. John Heitz III, H., 2002. Center for Science, Economics, and the Environment. The State of the Nation’s Ecosystem: Measuring the Lands, Waters, and Living Resources of the United States. Cambridge University Press, New York. Harrison, A.F., Latter, P.M., Walton, D.W.H., 1988. Cotton Strip Assay: an Index of Decomposition in Soils. Institute of Terrestrial Ecology, Grange-Over-Sands, Cumbria, England. Hopkinson, C.S., Schubauer, J.P., 1984. Static and dynamic aspects of nitrogen cycling in the salt marsh graminoid Spartina alterniflora. Ecology 65, 961–969. Jørgensen, S.E., 2002. Integration of Ecosystem Theories: A Pattern, third ed. Kluwer Academic Publ. Co., Dordrecht, The Netherlands, p. 432. Koch, M.S., Mendelssohn, I.A., 1989. Sulphide as a soil phytotoxin: differential responses in two marsh species. Journal of Ecology 77, 565–578. Koch, M.S., Mendelssohn, I.A., McKee, K.L., 1990. Mechanism for the hydrogen sulfide-induced growth limitation in wetland macrophytes. Limnology and Oceanography 35, 399–408. Kuhn, N.L., Mendelssohn, I.A., 1999. Halophyte sustainability and sea level rise: mechanisms of impact and possible solutions. In: Lieth, H., et al. (Eds.), Halophyte Uses in Different Climates. Backhuys Publishers, Leiden, The Netherlands, pp. 113–126. Lavorel, S.D., 1999. Ecological diversity and resilience of Mediterranean vegetation to disturbance. Diversity and Distributions 5, 3–13. Leps, J., Oshoronova´-Kosinova´, J., Rejma´nek, M., 1982. Community stability, complexity and species life history strategies. Vegetatio 54, 53–63. Matthaei, C.D., Uehlinger, U., Meyer, E.I., Frutiger, A., 1996. Recolonization by benthic invertebrates after experimental disturbance in a Swiss prealpine river. Freshwater Biology 35, 233–248. McKee, K.L., Mendelssohn, I.A., Materne, M.D., 2004. Acute salt marsh dieback in the Mississippi River deltaic plain: a drought-induced phenomenon? Global Ecology and Biogeography 13, 65–73. Mendelssohn, I.A., Kuhn, N.L., 1999. The effects of sediment addition on salt marsh vegetation and soil physicochemistry. In: Rozas, L.P., Nyman, J.A., Proffitt, C.E., Rabalais, N.N., Reed, D.J., Turner, R.E. (Eds.), Recent Research in Coastal Louisiana: Natural System Function
and Response to Human Influence. Louisiana Sea Grant College Program, Baton Rouge, LA, USA, pp. 55–61. Mendelssohn, I.A., Kuhn, N.L., 2003. Sediment subsidy: effects on soil-plant responses in a rapidly submerging coastal salt marsh. Ecological Engineering 21, 115–128. Mendelssohn, I.A., Morris, J.T., 2000. Eco-physiological constraints on the primary productivity of Spartina alterniflora. In: Weinstein, M.P., Kreeger, D.A. (Eds.), Concepts and Controversies of Tidal Marsh Ecology. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 59–80. Mitchell, R.J., Auld, M.H.D., LeDuc, M.G., Marrs, R.H., 2000. Ecosystem stability and resilience: a review of their relevance for the conservation management of lowland heaths. Perspectives in Plant Ecology, Evolution, and Systematics 3/2, 142–160. Patrı´cio, J., Salas, F., Pardal, M.A., Jørgensen, S.E., Marques, J.C., 2006. Ecological indicators performance during a recolonisation field experiment and its compliance with ecosystem theories. Ecological Indicators 6, 43–57. Penland, S., Ramsey, K.E., 1990. Relative sea level rise in Louisiana and the Gulf of Mexico 1908–1988. Journal of Coastal Research 6, 323–342. Rapport, D.J., Costanza, R., McMichael, A.J., 1998. Assessing ecosystem health. Trends in Ecology and Evolution 13, 397– 402. SAS Institute Inc., 1990. fourth ed. SAS/STAT User’s Guide, Version 6, vols. 1 and 2, fourth ed., SAS Institute Inc., Cary, NC, USA. Slocum, M.G., Mendelssohn, I.A., Kuhn, N.L., 2005. Effects of sediment-slurry enrichment on salt marsh rehabilitation: plant and soil responses over seven years. Estuaries 28, 519– 528. Thomas, G.W., 1982. Exchangeable cations. In: Page, A.L. (Ed.), Methods of Soil Analysis, Part 2. American Society of Agronomy, Madison, WI, USA, pp. 159–165. Turner, M.G., Romme, W.H., Tinker, D.B., 2003. Surprises and lessons from the 1988 Yellowstone fires. Frontiers in Ecology and the Environment 1, 351–358. Vitousek, P.M., Mooney, H.A., Lubchenco, J., Melillo, J.M., 1997. Human dominance of earth’s ecosystems. Science 277, 494–499. Walker, B.H., Langridge, J.L., McFarlane, F., 1997. Resilience of an Australian savanna grassland to selective and non-selective perturbations. Australian Journal of Ecology 22, 125–135. Webster, J.R., Waide, J.B., Pattern, B.C., 1975. Nutrient recycling and the stability of ecosystems. In: Mineral Cycling in Southeast Ecosystems. CONF 740513. National Technical Information Service, Springfield, VA, USA. Westman, W.E., 1978. Measuring the inertia and resilience of ecosystems. BioScience 28, 705–710. Whitford, W.G., Rapport, D.J., deSoyza, A.G., 1999. Using resistance and resilience measurements for ‘fitness’ tests in ecosystem health. Journal of Environmental Management 57, 21–29. Underwood, A.J., 1989. The analysis of stress in natural populations. Biological Journal of the Linnean Society 37, 51–78.
