Ecological Archives E095-235-A3

Jeffrey T. Wright, James E. Byers, Jayna L. DeVore, Erik E. Sotka. 2014. Engineering or food? mechanisms of facilitation by a habitat-forming invasive seaweed. Ecology 95:2699–2706. http://dx.doi.org/10.1890/14-0127.1

Appendix C. Detailed methods and GLM results for structural engineering effects of Gracilaria on amphipod survivorship.

Methods

Effects of Gracilaria on amphipod survivorship during low tide. Surface sediment collected from Priest Landing on September 9 was placed into twenty 80 mm diameter × 10 mm high petri dishes to a depth of ~ 5 mm and weighed. Gracilaria (8 g wet mass) was laid on top of the sediment in half of the petri dishes and kept ~ 5 mm from the edge of petri dish so amphipods were not able to crawl along the Gracilaria and escape. Five amphipods were pipetted onto the sediment at the edge of the Gracilaria and they immediately moved underneath the algal tissue. For dishes without Gracilaria the five amphipods were placed onto the sediment in the centre of each petri dish. At 2:15 pm, petri dishes were placed in a random array on the mudflat at Priest Landing and pushed into the sediment to a depth of 5 mm. The temperature of the sediment surface in petri dishes was recorded as 35°C.

After 1 hour, five petri dishes with Gracilaria and five without Gracilaria were collected and returned to the lab. The Gracilaria was removed and placed into a plastic bag. Each petri dish plus sediment was reweighed to determine loss of water from the sediment and then placed into a beaker of seawater for ~5–10 minutes which stimulated the movement of live amphipods in the sediment, allowing them to be enumerated. The Gracilaria was then rinsed to remove any amphipods and the sediment carefully searched. The total number of live and dead amphipods per petri dish was recorded. After 3 hours the remaining 10 petri dishes were collected and the process was repeated. Only one amphipod was not recovered from both experiments. Temperature of the sediment surface in the experimental petri dishes was not recorded to avoid disturbing the sediment and amphipods in them. Instead, temperature of the mudflat surface with and without Gracilaria (N = 8 of each) was measured approximately 20 m away from the experiment at 4.15 pm. We used a two-way factorial analysis of deviance to determine differences in Gammarus survivorship between Gracilaria (presence/absence) and time (1 and 3 hours). Because of large dispersion this GLM was performed with a quasi-binomial distribution with a logit-link and significance was tested against an F distribution (Crawley 2007). Differences in the percentage water loss from the sediment between petri dishes with and without Gracilaria at 1 and 3 hours were determined with a 2-factor ANOVA. The difference in surface temperature with and without Gracilaria was determined with a ­t test. We note that our estimates of temperature and water loss are relative between experimental treatments, and that sediments within petri dishes may vary slightly from the absolute rates of these parameters in natural sediments.

Effects of Gracilaria on predation of amphipods.

The shrimp Palaemonetes vulgaris and Panopeid mud crabs (Eurypanopeus depressus, Panopeus herbstii, and Dyspanopeus sayi) are common amphipod predators in this region (Van Dolah 1978, Nelson 1981) and were the most abundant mesopredators on these mudflats (J. DeVore unpublished data). Pilot studies confirmed they consumed Gammarus in the lab. Predation experiments were done in 6.15 L tubs (12 cm high × 34 cm long × 20 cm wide) which were placed into two large (~1000 L) shaded outdoor flow-through tanks. Tubs with different treatments were randomly positioned in the large tanks. These small tubs had openings (6 × 3 cm) cut into all four sides covered with 300 µm nitex mesh to allow for continuous water flow. The tops of the tubs were kept ~ 1.5 cm out of the water and were covered with translucent lids.

Sediment scraped from the top 2 cm of the mudflat was placed on the bottom of each tub and allowed to settle. A single piece of Spartina detritus ~ 4–5 cm in length was also placed in each tub to provide an additional food source for amphipods. 25 g wet weight Gracilaria was added to half the tanks and then 20 amphipods added to all tanks. After ~ 30 minutes, three predators (shrimp in experiment 1; crabs in experiment 2) were added at random to predator treatments and the lids placed on top of tubs. The mean total length of Palaemonetes was 20.75 ± 0.56 mm (mean ± SE) and the mean carapace width of the Panopeid mud crabs was 7.76 ± 0.19 mm (mean ± SE). Although ~ 90% of amphipods were Gammarus, there were at least two other species in the assemblage.For each experiment we identified 10% of all amphipods (N = 40) under a dissecting microscope which verified that 88% of amphipods in both experiments were Gammarus. Amphipods were added randomly to each replicate to ensure that each tub received a similar relative abundance of amphipod species.

Both experiments ran for 72 hours after which the tubs were carefully searched for amphipods. This involved sequentially removing the water, Gracilaria, Spartina and finally sediment from each tub, and rinsing them through a 500 µm sieve. We used GLMs in an analysis of deviance framework to determine differences in amphipod survivorship between Gracilaria (presence/absence) and predator (presence/absence) with a quasi-binomial distribution with a logit-link. Following the GLMs, we used multiple comparison tests done within the Gracilaria × Predator interaction using the multcomp package in R 2.12.1 (Hothorn et al. 2008) to determine whether survivorship of amphipods differed between treatments with and without Gracilaria when predators were present. For the Palaemonetes experiment, mesh on two tubs was loose at the end of the experiment (one each of the + Gracilaria / + shrimp and - Gracilaria / - shrimp treatments) so these tubs were removed from the analysis.

Table C1. Generalized Linear Models for Gammarus survivorship:A) the effect of Gracilaria (present vs. absent) and time (1 vs. 3 hours) at low tide, B) the effect of Gracilaria (present vs. absent) and the predatory shrimp Palaemonetes vulgaris (present vs. absent), and (C) the effect of Gracilaria (present vs. absent) and the predatory Panopeid mud crabs (present vs. absent). GLMs used quasi-binomial distributions.

Factor

df

Deviance

Residual df

Residual deviance

F

P

A) Gammarus survivorship at low tide

Null                     

 

 

19

86.689

 

 

Gracilaria

1

57.637       

18    

29.052  

42.205

< 0.001

Time

1  

1.345       

17    

27.707 

0.985

0.336

Habitat × Time

1

3.367

16

24.340

2.465

0.136

 

 

 

 

 

 

 

B) Gammarus survivorship vs. shrimp Palaemonetes

Null                     

 

 

17

299.549

 

 

Gracilaria

1

50.399       

16    

249.151  

16.608

0.001

Predator

1  

174.976       

15    

74.174  

57.659

< 0.001

Gracilaria x Predator

1

29.706

14

44.469

9.789

0.007

 

 

 

 

 

 

 

C) Gammarus survivorship vs. Panopeid mud crabs

Null                     

 

 

19

124.185

 

 

Gracilaria

1

27.365       

18    

96.820  

15.705

0.001

Predator

1  

57.448       

17    

39.373   

32.970

< 0.001

Gracilaria × Predator

1

7.054

16

32.318

4.049

0.061

 

Literature cited

Crawley, M. J. 2007. The R Book. First edition. John Wiley & Sons Ltd, Chichester, UK.

Hothorn, T., F. Bretz, and P. Westfall. 2008. Simultaneous inference in general parametric models. Biometrical Journal 50:346–363.

Nelson, W. G. 1981. Experimental studies of decapod and fish predation on seagrass macrobenthos. Marine Ecology Progress Series 5:141–149.

Van Dolah, R. F. 1978. Factors regulating the distribution and population dynamics of the amphipod Gammarus palustris in an intertidal salt marsh community. Ecological Monographs 48:191–217.


[Back to E095-235]