Assessment of the reliability of qQualitative Data Analysessamplingand Their Respective Reliabilityto characterize kelp forest communities
Title: 2 pts
Hardy et al.
The Kelp Forest Bible, 1st Edition, Vol. 1. (October 2011), pp. 1-10
The Kelp Forest Bible is currently published by Poseidon Publishing Inc.
Jordan… don’t use such small (10pt) font… use 11 or 12!
Abstract [0,0,4, 2, 4, 4] 14 pts
[First say what the question was… look at the writing guidelines for the components of this Abstract.] then… To answer these questions, wWe surveyed 28 species over 112 (???) thirty meter transects to determine the precision and accuracy of qualitative analysis. 5 vague interpretationscategories of general abundance levels were chosen for values of abundance and researchers observers were instructed not to discuss methods of interpreting these values prior to data collection. The data was WERE then collected and compared to determine the relative differences between individual qualitative analysis. (??) The results provided sufficient ranges in variance to conclude that qualitative data collection is not adequate for most types of researchestimates of abundance. The analysis did show that qualitative analyses do perform better when the abundance of populations are at either very low or very high and not between.extreme of the abundance descriptions. Qualitative analyses also prove to be more useful in the analysis to whether certain species are either absent or presentfor some species than others depending on such traits as….. Overall, quantitative data collection is preferential to qualitative in most circumstances and many factors need to be limited (???) when considering the use of qualitative data collection.
Introduction [4, 4, 4, 4, 0, 0, 0, 0, 4, 2, 2] 24 pts
The distribution and abundance of species are intimately interrelated and in many species and ecosystems, poorly understood. Understanding this relationship will provide insight into the biogeography, population genetics, evolution, and community dynamics of the species studied (Brown 1984). Understanding all of these concepts more comprehensively will broaden our understanding of which factors affect them and will provide insight on intraspecific and interspecific interactions, their strengths, and how the relationships between these interactions affect the community and the local players.
Abundance can be measured in either quantitative or qualitative methods and then analyzed to describe the distributions of the observed species. Depending on the conditions of the site described, the researcher may want to consider which method may be more accurate and the practicality of the chosen method. In general, qualitative analysis cannot compare to the accuracy and precision provided by quantitative analysis, however there are situations where quantitative methods may not be practical or even possible. Quantitative surveys provide detailed information that is not subjectively biased but require more time and resources, whereas qualitative surveys are fast and efficient but are subjective and can provide inaccurate data.
So, when is it appropriate to use qualitative analyses and can the data be accurate enough to provide conclusive data? Can a qualitative survey be repeated and confirmed? What qualitative descriptions are appropriate to provide reliable data and how can subjectivity be eliminated or reduced to an acceptable threshold to provide reliability?Species???
Develop this further by giving more description of the approach used… also describe the system (kelp forests) you used to conduct the study. We examined one method of qualitative analysis and calculated the percent difference between buddies to examine the reliability of the chosen method of analysis.
Give me some more citations in the Intro (at least 4).
Methods [2, 2, 0, 0, 0, 0, 0, 0, 2, 0, 4, 2, 4, 2, 0] 18 pts
Look at the guidelines… I need a general approach, then study system section, then separate sections for each of the questions and how you answered them with methods, predictions and analyses.
We used qualitative values to describe the abundance of 28 species at Hopkins Marine Life Refuge in Pacific Grove, California. [when?] There were 28 researchers involved in the collection of the data. The researchers were assigned a location along a 100m tape attached to the substrate at Hopkins reef. The locations were broken into 5m increments along the tape. There were 112 thirty meter transects performed (4 transects/buddy). A 30m tape was run out by 14 buddy pairs at their respective locations in two directions; the offshore transects ran 90 degrees due east followed by areciprocal transect back to the origin, and the onshore transectsran 270 degrees due west followed by a reciprocal transect back to the origin. The transects were not broken into area or volume bound sampling zones and were restricted only by the visibility seen from the 30m tape being run at each location. The researchers were asked to observe abundance of each of the 28 species (Fig. 1) and record the results after each transect. The abundance values were broken into 5 codes: 1 being absent, 2 being rare, 3 being present, 4 being common, and 5 being abundant. In addition, the researchers were asked to record the depth at the origin, the depth at the 30m mark on the offshore transects, the depth at the 30m mark on the onshore transects, and the visibility conditions during the survey. The buddy pairs were instructed not to discuss sampling techniques and code interpretation (i.e., what threshold of abundance do we assign to rare versus common). The data was WERE then collected and compared to determine first the precision of the method and then the accuracy. Variance component analysis was conducted by first measuring the overall percent variance among 3 sources (depth, buddy, and meter) and then the same sources were measured after being broken into 3 categories of taxa (algae, fish, and invertebrates). The mean abundance, relative difference between buddy assessment of abundance, and percent difference of disagreement to whether a species was absent or present was calculated for each of the 28 species. The difference between buddy pairs was then compared as a function of mean abundance of species.
Fig. 1 – This table describes the 28 species observed in the study. It categorizes the species by taxa, species, the code used for the species in the results, and the common name for each species.
Results
The depth of each transect accounted for 23% of variance. The difference in buddy qualitative methods accounted for 37% of variance. The location of transects along the meter tape accounted for 40% of the variance.
Fig. 2 – Percentage of variance is shown as a function of the source causing the variance.
In algae, 37% variance was caused by buddy qualitative methods, 10% was caused by depth of transect, and 53% was caused by transect location. In fish, 17% of variance was caused by buddy qualitative methods, 23% was caused by depth of transect, and 60% was caused by location of transect. In invertebrates, 50% of variance was caused by buddy qualitative methods, 50% was caused by depth of transect, and 0% was caused by transect location.
Fig. 3 – Percentage of variance is shown as a function of the source causing the variance. Percent of variance is now broken into three categories of taxa (algae, fishes, and invertebrates).
Fig. 4 – The mean abundance of each species is measured on a scale from 1-5 (1-absent, 2-rare, 3-present, 4-common, 5-abundant). The species are separated along the x-axis by the codes used to describe them (Fig. 1).
The mean abundance was measured by taking the mean of the total value found by all buddy teams for each species.
Fishes
Damalichthysvacca, Embiotocajacksoni, and Oxylebiuspictus had a mean abundance of 1.8. Embiotocalateralis, Sebastes atrovirens, and Sebastes mystinus had a mean abundance of 1.7. Hexagrammos decagrammus had a mean abundance of 1.6. Sebastes carnatus had a mean abundance of 1.3 and Sebastes chrysomelas had a mean abundance of 1.2.
Algae
Macrocystis pyrifera had a mean abundance of 3.8. Cystoseira osmundacea had a mean abundance of 3.7. Dictyoneurumcalifornicum had a mean abundance of 2.2. Dictyoneurumreticulatum and Chondracanthuscorymbiferus had a mean abundance of 2.1 andPhyllospadix spp. had a mean abundance of 1.3.
Invertebrates
Patiriaminiata had a mean abundance of 4.2. Pisastergiganteus had a mean abundance of 3.0. Balanophylliaelegans had a mean abundance of 2.8. Calliostomaligatum and Pachycerianthusfimbriatus had a mean abundance of 2.5. Tethyaaurantia had a mean abundance of 2.4. Pisasterbrevispinus had a mean abundance of 1.7. Pycnopodiahelianthoides had a mean abundance of 1.6. Urticinapicivora had a mean abundance of 1.5. Loxorhynchusgrandis had a mean abundance of 1.4. Urticinalofotensis had a mean abundance of 1.1. Haliotisrufescens and Strongylocentrotusfransiscanus had a mean abundance of 1.0.
Fig. 5 – The percent difference between values determined by buddies broken down by each species observed.
Fishes
Damalichthysvacca, Sebastes mystinus, andEmbiotocajacksoni had a relative difference of 33%.Embiotocalateralis had the highest relative difference of 45%. Hexagrammos decagrammus had a relative difference of 40%.Oxylebiuspictus had a relative difference of 43%. Sebastes atrovirens had a relative difference of 28%. Sebastes carnatus had a relative difference of 25% and Sebastes chrysomelas had the lowest relative difference of 12%.
Algae
Cystoseira osmundacea had a relative difference of 25%. Dictyoneurumcalifornicum had a relative difference of 45%. Dictyoneurumreticulatum had the highest relative difference of 50%. Chondracanthuscorymbiferus had a relative difference of 28%. Macrocystis pyrifera had the lowest relative difference of 20% and Phyllospadix spp. had a relative difference of 30%.
Invertebrates
Balanophylliaelegans had the highest relative difference of 58%.Calliostomaligatum had a relative difference of 51%.Haliotisrufescens had a relative difference of 9%.Loxorhynchusgrandis had a relative difference of 35%.Pachycerianthusfimbriatus had a relative difference of 54%.Patiriaminiata had a relative difference of 22%.Pisasterbrevispinus had a relative difference of 41%.Pisastergiganteus andTethyaaurantia had relative differences of 48%. Pycnopodiahelianthoides had a relative difference of 21%.Strongylocentrotusfransiscanus had a relative difference of 10%. Urticinalofotensis had the lowest relative difference of 8% andUrticinapicivora had a relative difference of 17%.
Fig. 6 – Shows the percent of 14 buddy pairs that disagreed whether each species was either absent or present (1-absent, 2,3,4,5-present).
Fishes
Damalichthysvacca had a percent disagreement of 34% between buddies.Embiotocajacksonihad a percent disagreement of 27% between buddies.Embiotocalateralishad a percent disagreement of 43% between buddies. Hexagrammos decagrammushad a percent disagreement of 38% between buddies.Oxylebiuspictushad a percent disagreement of 40% between buddies. Sebastes atrovirens and Sebastes mystinushad percent disagreements of 25% between buddies. Sebastes carnatushad a percent disagreement of 37% between buddies and Sebastes chrysomelashad a percent disagreement of 10% between buddies.
Algae
Cystoseira osmundaceahad a percent disagreement of 9% between buddies.Dictyoneurumcalifornicumhad a percent disagreement of 41% between buddies.Dictyoneurumreticulatumhad a percent disagreement of 47% between buddies.Chondracanthuscorymbiferushad a percent disagreement of 20% between buddies. Macrocystis pyriferahad no disagreement between buddies andPhyllospadix spp. had a percent disagreement of 36% between buddies.
Invertebrates
Balanophylliaelegans, Loxorhynchusgrandis, and Pisasterbrevispinus had percent disagreements of 39% between buddies.Calliostomaligatumhad a percent disagreement of 30% between buddies.Haliotisrufescenshad a percent disagreement of 10% between buddies. Pachycerianthusfimbriatushad a percent disagreement of 47% between buddies.Patiriaminiatahad no disagreement between buddies.Pisastergiganteushad a percent disagreement of 29% between buddies.Pycnopodiahelianthoideshad a percent disagreement of 26% between buddies.Strongylocentrotusfransiscanusand Urticinapicivorahad a percent disagreement of 14% between buddies.Tethyaaurantiahad a percent disagreement of 34% between buddies andUrticinalofotensis had a percent disagreement of 7% between buddies.
Fig. 7 – The percent difference between buddies as a function of the values assigned to describe the abundance of species.
The mean abundance differed between each categorical value of abundance for overall data. The lower values tended to have a larger percent of difference among buddies (between 50-80% for abundances <3) and a lesser percent of difference for higher values (between 10-49% for abundances >3).
Discussion
All results showed a wide range of opinions about the degree of abundance. The variance component analysis (Fig. 2) shows that the highest degree of variability was among location of transect. The second highest contributing source was buddy qualitative analysis. The buddy component needs to be much lower than 37% for results to be considered precise and held to an even higher standard for results to be considered accurate. This result shows that qualitative analysis for 28 different species is not reliable.
When broken into taxa we see that the buddy component drops to a 17% difference for fish. This result is surprising because of the inherent difficulty of fish surveys. I would expect the difference to be higher, however the lower difference can be accounted for the complete absence of most fish species. Many fish are cryptic or simply evasive and may not have been noticed or present during the survey. This would cause a higher agreement within the buddy component because many of the fish would be recorded as absent. Furthermore, 2 of the 4 transects were completed after reeling the tape out 30 meters. During the reciprocal survey back toward the origin, we can expect that most of the fish that were within the survey area have left the area, resulting in more absent responses. This causes the data to appear to be more valid but in reality, fish may be some of the worst candidates for qualitative surveys. This conclusion is supported with the percent disagreement upon the absence/presence of fish species (Fig. 6). The fish have a consistently high rate of disagreement among buddies across all species aside from the one outlier, Sebastes chrysomelas. Sebastes chrysomelas is probably the most distinct species among all those on the species list. The distinct black and yellow pattern present in all individuals makes identifying this species a relatively simple task. However, one must wonder how the data may have changed if Sebastes nebulosus had been added to the survey sheet. Sebastes nebulosus shares a very similar coloration and it would take a more trained eye to distinguish between the species.
Taking a qualitative analysis to the level of presence versus absence is the most basic and probably the most reliable way to perform any type of qualitative survey. The mobility of fish cause accurate identification to be minimal. The visibility also plays an integral part in fish surveys. To get the most accurate result from a fish survey, I would suggest targeting one or two species per surveyor at any time so that they can focus their search image and conducting the surveys during times of relatively good visibility.
Algae were very surprising within the variance component analysis. I would have expected algae to be the lowest in variance considering there are few species in the area, they are all sessile organisms, and most of the species are very distinct from each other. Algae had a 37% degree of variance among buddies. The variance may be explained by the degree of experience individuals had diving in kelp forest environments. A Macrocystis alga may seem abundant to one buddy who had not been diving in dense Macrocystis environments where the abundance may seem relatively rare to another individual who had experience diving in forests teeming with Macrocystis. In contrast to the high variance of perception of algal abundance, when the data was compared against the presence versus absence of algae, the data seems to be the most valuable of all collected. Macrocystis pyrifera is one of the only 2 species among the data that received a 0% variance in the presence versus absence analysis. This is most likely due to the integral part Macrocystis pyrifera plays in a kelp forest community. The presence of Macrocystis pyrifera is difficult to miss. It is a distinctive canopy alga that is very abundant in the forests it is observed in. I would expect the same result for Nereocystis luetkeana in kelp forests dominated by this species based on the same conditions. Qualitative analysis among algae should prove most efficient if the researchers all share similar diving experience and the qualitative codes (i.e., common versus rare) should be discussed beforehand and a threshold of abundance should be agreed upon.
Invertebrates have a very wide range of results in our data. 50% of variance can be attributed to buddy analysis, 50% of variance can be attributed to depth, and 0% of variance can be attributed to the location of transect. The 0% variance among transects is a surprising result because of the high richness of invertebrates inherent to the kelp forest community. Perhaps this high richness is the direct cause of the low variance. It is possible that a 30 meter transect is such a large distance that all invertebrates on the data sheet can be expected to be encountered. This result shows that it is highly likely that no transects were assigned exclusively within a sand bed. In healthy kelp forest it is likely to find a large majority of the 13 observed species upon one rock or within a small area. In contrast the relative difference between buddies appears to be higher overall compared to the fish and algae. This again can be contributed to the individual’s experience in diving with invertebrates. Some diver’s may have better search images for invertebrates and may consider high densities of certain species to be of a lesser relevance than another researcher. For instance 20 Patiriaminiata in one 30 meter transect may seem to be only common to one researcher where another may consider this to be abundant simply because the species seems to outnumber any other observed species. When compared in the presence versus absence comparison, Patiriaminiata is the only species that resulted in a 0% variance among buddies. This is most likely contributed to the common nature of the species. It is also a very distinct species, however I would like to see if the 0% variance would hold up if Dermasteriasimbricata had been on the species list. Dermasteriasimbricata shares a similar morphology and can be easily confused when observed from a distance. Qualitative data pertaining to invertebrates should be most efficient with very experienced researchers and the amount of species being observed by any one researcher should be kept to a minimum.
All qualitative analysis would produce better results if the target species are kept to a minimum. The search images any one observer can keep track of are limited and even if all species are observed, the relative abundance can be easily skewed depending on the abundances of other species in the area and how long the diver must retain the information they are processing. Quantitative data would be best to describe the abundance of species through time because the quantitative analysis allows a more precise determination of density. In situations where quantitative data would not be practical, qualitative can be accurate as long as strict thresholds are observed. Qualitative analyses would work best with species that are on either extreme of high or low abundance. It would also be best to assemble a dive team with similar experience and to conduct many practice dives in the area planning to be observed so that the researchers can get a feel for what abundant versus rare is for that community.