The terms survey and monitoring are often used interchangeably when collecting data on the marine environment. More worryingly there is sometimes a blurring of the differences between the aims and methods required for descriptive surveys and data collection as part of a time-series monitoring programme.
In general, the approach to benthic survey differs from that taken to monitoring in a couple of important aspects. In descriptive survey, more emphasis is put on identifying what organisms and habitats are present rather than precisely how many of them are there. Conversely, monitoring often does not give full descriptions of sites, it may only look at a sub-set of organisms, but requires greater precision in recording the number of individuals or colonies or colonies present, or more precise measurement of the condition of the organisms present. The reasons are fairly obvious; with monitoring you are concerned about recording change in numbers, so your numbers need to be pretty accurate and you need to be pretty confident about the identities of what you are recording. This means that we have to be fairly careful when designing or selecting the format in which data is collected. For example, the UK’s Joint Nature Conservation Committee’s Marine Nature Conservation Review’s recording forms are based of the SACFOR Abundance Scales (apologies for all the acronyms). These SACFOR* scales are widely used for marine survey work around the UK today. They have the advantage that they are well known, widely accepted and can be applied to all marine habitats and all marine marcoflora and fauna. They are very useful for descriptive surveys; they give a good feel for the composition of species assemblages and with experience broad comparisons can be made between different sites. Unfortunately they are also sometimes used for time-series monitoring, something for which they are pretty useless. As they use a logarithmic scale each abundance category is an order of magnitude up or down from the next, thus you need in the region of a ten-fold change in abundance to register as change on the scale. For most species change is abundance will be very obvious long before a ten-fold change in abundance occurs, thus recording only SACFOR abundance values will mean that quite large impacts (e.g. a 50% reduction in a key species abundance, or the doubling in numbers of an invasive species) may go un-noticed. This scenario happens all to easily, especially where one organisation is contracted to undertake repeat monitoring and compare data with that collected by a different organisation (something I have often had to do) or when different staff undertake monitoring on different years.
A corollary of this is that you also need a good idea of what your margin of error is and what are your sources of error. These sources of error are particularly important to know if they are variable or intermittent. Again the reasons are fairly self-evident. If you have sources of error that affect the data they need to be identified and recorded if erroneous records of change or false conclusions are to be avoided. A good example of this in diver or remote camera recording is underwater visibility. The waters off Southern England are prone to strong plankton blooms during the summer months. These blooms vary in timing and duration. Sometimes they arrive in late April and linger for months; sometimes the do not arrive until mid-May and disappear after a few weeks. They also vary in intensity and distribution. When the plankton is thick visibility can be 0.5 of a metre or less, often the plankton occurs in patches, so that visibility is less than a metre at one location but several metres only a few miles away. This patchiness can vary from day to day and from hour to hour as the tides sweep in water from upstream locations. Similarly storms and string tides can lift sediment in to the water column, similarly reducing near seabed visibility to a fraction of what it was days before. This can make visual comparisons between different points in time extra-ordinarily difficult. This is particularly true of comparisons between photographs taken as part of a time series. When the visibility is reduced through plankton blooms, strong tides or following poor weather this can dramatically reduce the number of individuals counted within a fixed area when no change in numbers has actually occurred. Thus it is vitally important for monitoring studies that the raw data (i.e. photographs or log sheets with condition records) and not simply numerical count data is available to those tasked with interpretation of the data.
Because we need greater precision and numerical accuracy for monitoring there are differences in the appropriate methods. Video, either diver operated or remote, can be really useful for broad-scale survey as it collects a lot of spatial data quickly and cheaply and can be very useful for identifying habitats and some conspicuous species or flora/fauna types (e.g. for identifying biotopes as hydroid/bryozoan turf or red algal turf or kelp forest). It can also useful for counting larger, conspicuous and widely spaced individuals (e.g. estimating densities of mature seafan colonies), though stills photography sampling or mosaics are normally a much a better option. Video is rarely suitable for monitoring smaller faunal turf species (such as sponges, soft corals, anemones, hydroids, tunicates etc.) as, although quality is steadily improving, video still does not have the image resolution for accurate identification and accurate counts. This is not to say it will never see some of them; rather it may possibly see some but exactly how many in relation to how many are actually there will vary considerably so the data generated will be unreliable.
Species for monitoring.
For any given study we will select target species for monitoring based on a criteria such as known or expected sensitivity to the variable (e.g benthic filter feeding organisms may have a know or presumed sensitivity to increased suspended particulates and sedimentation rates due to nearby dredging or spoil dumping). However, there are some fundamental criteria that apply to nearly all monitoring studies.
- We must be able to find the species using the selected methodology (e.g. if using remote towed or drop cameras, species that tend to live of hide in crevices or under rock overhangs are generally unsuitable because they are only likely to be recorded by chance and so numbers are likely to be unreliable)
- We must be able to accurately identify the selected species using the selected methodology;
- The species must be evenly distributed across the habitats in question (e.g rare species confined to a small area within the total study area are unlikely to yield useful data, especially where a study involves treatment and control areas).
- The species must occur in numbers sufficient to generate statistically usable data;
- The methodology employed must be able to accurately count the selected species otherwise error or bias will occur.