Is this pseudo-replication?

Good morning,

I am trying to analyze data from a study that was conducted in 1999-2000. Some of the issues are that 1. I wasn't even working with the organization when the study was conducted, so I am trying to understand the scope of study from project documents and personal communication with other staff that were present at the time; 2. while I did take several statistics classes (liner regression, multivariate regression, and experimental design), that was several years ago, so I'm a bit rusty; and 3. the data was collected without thinking about how it was going to be analyzed, so the study is a bit messy, and I'm not sure where to start.

Okay, so with that, here is a brief synopsis of the study:

1. The purpose of this study was to understand the relationship between habitat characteristics and a single shorebird species' breeding success on an island in the mid-Atlantic. There were several research questions, but the one I am focusing on at this time is describing the habitat characteristics at different elevation classes. I would then to compare the elevation classes and determine if there is any difference in there characteristics. The main hypothesis for this question was that the elevation classes would be significantly different at alpha = 0.1.

2. For describing habitat, the study looked at several variables: vegetation alliance, sediment characteristic, surface soil moisture, ambient wind, surface temperature, prey abundance, overwash activity, wrack description, and elevation class.

Basically, I here's what I would like to be able to say: at Elevation Class 1 (elavations ranging from 0.5 to 1.0 m above sea level) the vegetation alliance is ___, the sediment characteristic is _____, etc.

Now the island is sort-of like a long rectangle, running in a north-south direction. Every 1/2 Km a transect was established and run from the ocean side to the bay side of the island. 18 transects were established on a portion of the island where 95% of the shorebirds nests and raise their offspring. A GPS was used to determine elevation every 10 m along this transect (this is actually done for another study, but was also used in this study). Then elevation was broken into several 1/2 meter classes. There ended up being eight different elevation classes (EC). One thing I noticed is that the classes made no distinction between the ocean side of the island and the bay side of the island. Thus, along the transect, elevation class 2 could be found on the east side and west side of the dune line.

Because the study was focused on one particular shorebird that only foraged in sparsely vegetated areas, this influenced some of the stopping rules. For example, once the island transects had been divided up into elevation class, (i.e. the island was 400m wide, at transect 7 there was 75m in EC 1, 50m in EC 2, 25m in EC 3, 75m in EC 4, 150m in EC 5, and 25m in EC 6), one stopping rule was that in order for a transect segment in a particular elevation class to be sampled, it had to have a minimum 30m length of uninterrupted sparsely vegetated habitat. (The density of vegetation was described in the study for sparsely vegetated habitats.) Thus, out of 18 transects, only 12 of them may have suitable EC 4 segments. Then from those 12, 6 were randomly selected to be sampled.

And now, if you are still with me - here comes the pseudo-replication question: Within each of the six transect segments, 5 sampling points were laid out. If the segment was only 30m long, then they were every 5m apart (with a 5m buffer on each end of the segment). If the segment was 100m long, then the sampling points were every 20m apart. So in your opinion, would these sampling points, which for the original researcher, were the replicates, actually be the pseudo-replicates?

I know that I still haven't answered question 3, but I think I have given you all enough to digest for now.

On a final note, while I have had some statistics courses, this is not my strong point. I cannot count the number of similar problems that arise in many biological studies. Basically, it comes down to rushing out into the field to collect data without formulating a clear experimental design - even in this case where there were no tests being conducted, the study could have benefited from someone asking up front - how is the data going to be analyzed. I know the saying garbage in garbage out, but I am hoping is that something useful can come from this study. Any thoughts, comments, or suggestions on how to proceed would be greatly appreciated.

Cheers (and thanks),
Last edited:


TS Contributor
About a year ago I was doing some Google searches on pseudoreplication because it came up as a work-related question - I was amazed at the number of hits that included ecology, biology, zoology, field studies, etc. - probably at least 75-80%

Anyway, if you consider that by definition, a replication is an independent repeat of a treatment or treatment combination (the critical term here is independent), then anything that falls short of that standard should be considered a pseudoreplication.

If the researcher wants to call it a replication and has specific reasons for doing so, then I would try to be a bit pragmatic - however, it should be noted in the report that they're not "true" replicates.