Sertic 1
Maya Sertic
Pete Raimondi, Mark Carr
Kelp Forest Ecology
October 11, 2011
Reliability of Qualitative Sampling for Surveying Species Abundance in a Subtidal Environment
Introduction
Species distribution is an important characteristic in describing and comparing ecosystems, and in recognizing changes in ecosystems over time. Every species holds a specific niche within its community, and disturbances can change these niches and displace or eliminate entire populations. One way of detecting environmental changes and their impacts on a given ecosystem is to survey communities over time and evaluate changes in species abundance (Holbrook et al. 1997).
Monitoring population abundance in marine subtidal ecosystems is a challenge because divers are limited by the amount of time they can stay underwater. While quantitative sampling aims towards objectivity and tends to produce reliable data, it takes a lot of time.By contrast, qualitative sampling helps to save time when surveying subtidal communities, while still providing information on population abundance and distribution. Institutions such as the Reef Environmental Education Foundation (REEF, organizer of the Great Annual Fish Count) train divers to take qualitative data on species abundances while diving. The surveys are compiled for use in monitoring the conditions of marine biota over time.The data have been used for many purposes, including scientific research (REEF 2007),assessing the effects of humans on marine fish populations (Stallings 2009, Semmens et al. 2004), and managing marine fisheries with regard to changing fish populations (Biological Review Team NWFSC 2009 unpublished).
Although qualitative sampling methods have a huge advantage in that much more data can be collected in a given time period, the resulting data relies heavily on the opinion of the diver. Qualitative sampling runs the risk of being biased due to the observer’s previous diving experience, depth of background knowledge about each species, style of sampling, or personal interests. Even so, tests have shown that qualitative sampling is an appropriate method for monitoring certain ecosystems, such as shallow streams (Lenat 1988). We wanted to test the reliability of data collected by qualitative sampling in subtidal communities, and to determine whether such methods are reasonable for marine environments.We examinedwhether there is a difference between individual divers in qualitatively determining species abundances, and whether some species are better candidates for qualitative sampling than others.
Site
We conducted our study in the kelp forest at Hopkins Marine Reserve, in Pacific Grove, California. Our study site was dominated by Macrocystis pyrifera, which provides habitat for many marine invertebrates and fishes (Watanabe 1984).
Methods
28 divers with training in scientific diving were paired haphazardly into buddy pairs. Each diver individually collected data on the abundance of 28 fish, algae, and invertebrate speciesby assigning a score of 1 to 5 for each species (1=absent, 2=rare, 3=present, 4=common, 5=abundant). Divers used their individual experience to determine the abundance of each species, and theydid not discuss methods for determining species abundance with their buddies. There were no guidelines on how to search for the organisms other than to stay next to the transect line and buddy.
We conducted our dives in the late morning on September 27, 2011. The weather was clear and sunny, but the visibility under the kelp canopy was less than 3m throughout the site. The 14 buddy pairs were spaced 5m apart from one another along a main cable running north-south through the kelp forest (Fig. 1). Each buddy pair ran two 30m transects, one at 90° west (deep transect), and one 270° east (shallow transect) for a total of 28 transects.While running out the meter tape, both buddies in each buddy team individually collected data on the abundances of the 28 species (9 fish species, 6 algae species, and 13 invertebrate species). We collected qualitative abundance data for all of the species while reeling out the meter tape, and then again as we reeled the meter tape back in (so that each transect was surveyed twice). Both buddies in each buddy pair collected data at the same location at the same time. We also noted the depth at either end of each transect. As we reeled the measuring tape back in, we used the meter marks on the tape to determine the visibility.
We used Variance Components Analysis to assess the relative contributions of three potential sources of variance to our estimates of the abundances of fish, algae, and invertebrate species. We distributed the total variance among spatial patterns (transect depth and microsite in which the transect was laid out) and differences between individuals within buddy pairs.
Results
Macrocystis and Cystoseira were the most common algae found at Hopkins, while Phyllospadix species were very rare. The invertebrates we sampled ranged in abundance from absent to common, while all the fish species scored an average below 2 (“rare”). The standard error for the sampled abundance data for all species was very high, so we removed the effects of differences between buddies from the standard error to better focus on the abundance data (Fig. 2). Our variance components analysis showed that much of the variance was accounted for by differences in microsites along the main cable. However, almost 40% of the variance was attributed to differences in scoring between buddies (Fig. 3).
For most of the species surveyed, the relative difference in scoring between buddies was high, with several cases in which there was more than 50% disagreement. For a few species, such as Haliotis rufescens, Strongylocentrotus franciscanus, Urticina lofotensis, and Sebastes chrysomelas, buddies tended to agree on level of abundance (about 10% difference in scoring between buddies). There were no cases for which there was 100% agreement between buddies (Fig. 4). When we grouped all of the sampled abundance scores greater than 1 into a single category, so that each species was scored as either “present” or “absent,” the disagreement between buddies dropped below 50% for all species. Macrocystis pyrifera and Asterina miniata were the only 2 species for which there was agreement between buddies for all 28 transects. The disagreement between buddies for the presence of some fish species increased (Fig. 5).
Abundance scores of 1 and 5 tended to have higher rates of agreement between buddies than the intermediate scores. Species that were assigned lower mean abundances had the highest rates of disagreement between buddies (Fig. 6).
Discussion
Our test of qualitative sampling methods gave imprecise abundance data for most of the species sampled, suggesting that qualitative methods are unreliable for estimating species abundances in kelp forest habitats. Scientific research is most reliable when there is no effect of human error on the data. This is not only important for increasing the precision of the data itself, but also for analyzing the sources of variance. When a high portion of the total variance is attributed to human error, it obscures the effects of the environment on the data. Because about 40% of the variance in our data was attributed to differences in sampling between buddies, the variance in abundances of species due to depth and microsite was not as pronounced. Consequently, the importance of these environmental factors in affecting population distributions is uncertain. Many scientific studies use population distributions to test the effects of environmental changes on ecosystems (eg. Holbrook et al. 1997), and concealing the effects of environmental factors with human error may result in false conclusions.
There was a very high rate of disagreement in the scoring of species abundances between buddies in our study. We found that species which were absent and species which were very abundant tended to have a lower difference in scoring between dive buddies. Perhaps the terms “absent” and “abundant” both have a more specific meaning to people than “rare”, “present”, and “common.” Species with a low sampled abundance had an especially high rate of disagreement between buddies, suggesting that divers with different backgrounds have different definitions of “rare” for different species. For example, some divers may rate all species on the same scoring system, where 1 observed individual means the species is “rare”, seeing more than 1 individual means the species is “present”, seeing more than 5 individuals means the species is “common”, and seeing more than 10 individuals means the species is “abundant.” Other divers might use different scoring systems based species’ ecological roles (predators tend to have lower abundances than herbivores) or life histories. Divers with diving experience from different regions may also use different qualitative abundance ratings to describe the same population. Blue rockfish are not nearly as abundant south of Point Conception as they are in the north (Key et al. 2007), so observing 20 blue rockfish may seem “abundant” to a diver from the south, but only “present” to a diver from the north.
We controlled for differences in diver backgrounds by grouping the collected data into “present” and “absent” categories. Although differences between buddies did drop considerably, for most species there was still disagreement between buddies. This difference between buddies can be attributed to two variables: sampling style and ability to correctly identify species. Problems with species identification can be removed relatively simply through education, and institutions like REEF do provide training seminars which teach basic species identification to divers (REEF 2007).
Sampling style may vary from person to person based on interests in certain taxa or attention to detail. We purposely did not create a set of guidelines for divers to follow while sampling, which may have affected what divers observed. While species like Macrocystis pyrifera and Asterina miniata were “common” at Hopkins (Fig. 2) and are relatively conspicuous, other species may be small (Balanophyllia elegans), cryptic (Oxylebius pictus), drably colored (Pachycerianthus fimbriatus), hidden (Loxorhynchus grandis), or motile (most fish). Moreover, while most invertebrates are found on substrate, many fish are in the water column, making it challenging to account for all species in a given area at once (especially if some species are motile and flee from divers). Collecting data only on species that share a similar ecological niche would focus attention on part of the reef and minimize the risk of not seeing some species.
Divers may have special interests in certain taxa, and might subconsciously pay more attention to one section of a reef than another, biasing their results.Thus while one individual may spend more time searching in cracks and crevices for hidden invertebrates, another may be peering out into the water column and observing fish. Creating sampling guidelines for divers to follow would likely minimize differences in sampling between buddies for presence/absence data.
Although our study suggests that qualitative sampling is unreliable, the benefit of being able to sample quickly in a subtidal environment is huge. Thus it is worthwhile to look into restructuring our methods to minimize human error. Individual studies may control for human error by minimizing the number of divers collecting data, thus standardizing the human error across all surveys (Holbrook et al. 1997). However, this is only realistic for small projects. Large regional surveys like those arranged by REEF depend on a large pool of surveyors. Major adjustments to the methods for these surveys may help decrease variance due to human error. Our recommended changes to the methods include: 1) simplifying the scoring system and defining it for taxa with different life histories, 2) searching for species that occupy a similar habitat during a given sampling period, 3) standardizing sampling techniques among divers, 4) evaluating diver abilities to correctly identify species as well as their familiarity with the region. The true effect that these changes will have on variance due to human error needs to be tested before it is accepted. In any case, it is very likely that human error cannot be completely removed from qualitative sampling. Thus, this method should not be relied on for detailed investigations or for recognizing subtle changes in the environment, especially in sensitive ecosystems. Ecosystems that can be greatly affected by small environmental changes should be monitored using quantitative methodsto minimize noise and to give people a chance to respond before those ecosystems are irreversibly damaged.
Literature Cited
Biological Review Team NWFSC. 2009. Preliminary scientific conclusions of the review of the status of 5 species of rockfish. (unpublished?)
Holbrook SJ, JS Russell, JS Stephens Jr. 1997. “Changes in an assemblage or temperate reef fishes associated with a climate shift.” Ecological Applications. 7(4): 1299-1310.
Key M, AD MacCall, J Field, D Aseltine-Neilson, K Lynn. “The 2007 Assessment of Blue Rockfish (Sebastes mystinus) in California. In: Status of the Pacific Coast Groundfish Fishery Through 2007, Stock Assessment and Fishery Evaluation: Sticj Assessments and Rebuilding Analyses Pacific Fishery Management Council: Portland, OR.
Lenat DR. 1988. “Water quality assessment of streams using a qualitative collection methods for benthic macroinvertebrates.” Journal of the North American Benthological Society. 7(3): 222-233.
REEF. 2007. Reef Environmental Education Foundation. 10 October 2011. <
Semmens BX, ER Buhle, AK Salomon, CV Pattengill-Semmens. 2004. “A hotspot of non-native marine fishes: evidence for the aquarium trade as an invasion pathway.” Marine Ecology Progress Series. 266: 239-244.
Stallings, C. 2009. “Fishery-independent data reveal negative effect of human population density on Caribbean predatory fish communities.” PLoS ONE. 4(5): e5333. Doi: 10.1371/journal.pone.0005333.
Watanabe JM. 1984. “The influence of recruitment, competition, and benthic predation on spatial distributions of three species of kelp forest gastropods (Trochidae: Tegula).” Ecology. 65(3): 920-936.
Results (25)
__4__/4 Figure legends Accurate
__4__/4 Figure Legends well composed (complete and concise)
__5__/5 Results organized according to questions
__4__/4 Graphs presented in a logical order, case made for the order
__4__/4 Grammar, sentence structure and spelling
__4__/4 Clarity and conciseness of writing
Discussion (25)
____/9 How well did they answer the questions they present in the Intro?
1)__3__/3 Discuss the results from the specific to the general.
2)__3__/3 Do these results surprise you? In other words, is the qualitative method more or less reliable than you thought it would be, and do you think that degree of reliability (which can be assessed based on relative difference between buddies) implies anything about accuracy?
3)__3__/3 Do you think the qualitative sampling approach is appropriate for describing trends of species abundances through time? Explain your answer
__3__/3 Grammar and Spelling
__2__/2 General Thoughtfulness
__2__/3 Clarity and conciseness
__4__/5 Organization of discussion
__3__/3 Context and Bigger Picture
General Notes: Excellent job. I don’t have many comments for you. Good job on figure legends. Your discussion hits on all the main points and I really like the way you finished up with recommendations on how to improve the process.