IMPROVING ANALYSIS AND SYNTHESIS OF ANIMAL POPULATION STUDIES

• Sponsoring Institution: National Institute of Food and Agriculture
• Project Status: TERMINATED
• Funding Source: HATCH
• Reporting Frequency: Annual
• Accession No.: 0221318
• Project No.: TEX09359
• Project Start Date: Feb 1, 2010
• Project End Date: Jan 31, 2015

Project Director
Fujiwara, M.

Recipient Organization
TEXAS A&M UNIVERSITY
750 AGRONOMY RD STE 2701
COLLEGE STATION,TX 77843-0001

Performing Department
Wildlife & Fisheries Sciences

Non Technical Summary
Understanding how population abundance fluctuates is a fundamental question in ecology. The answer to this question will provide important information for the management of natural populations under changing environmental conditions. Ecologists have developed a plethora of ecological theories that can potentially explain observed population dynamics. Population management often relies on them. However, the concepts typically are not validated with actual data. Another major factor affecting the way population abundance fluctuates is environmental variation, which affects the biology that underlies various population parameters. Unfortunately, researchers have had only limited success in linking environmental change to observed fluctuations in population abundance. This limitation exists even though researchers have access to thousands of population datasets. Lack of success in explaining fluctuations in fish and wildlife population abundance partly stems from our inappropriate use of statistical methods. Currently, the most common population data analyses are simple correlation and related regression-type analyses, but these methods alone are not sufficient. Another type of problem in ecological studies is a mismatch between an ecological question at hand and the type of data collected. The most common population data collected are the time series of individual counts. However, it is sometimes more informative to collect capture-recapture type data depending on the particular issue at hand and/or type of organism we study. Because it is often not feasible to switch from one type of data to another in the middle of a study, the decision has to be made in an early stage of the study. However, there currently does not exist a comprehensive guideline for selecting data types in ecology. The objective of the proposed project is to develop a comprehensive set of guidelines for the analysis and collection of population datasets in both applied and basic ecological research. This will be achieved by applying a series of statistical methods (e.g. multivariate time series and capture recapture statistics) to simulated data sets as well as existing data sets available from Texas and other parts of the US. The guidelines are expected to substantially improve the way population studies are conducted, enhance our understanding of ecological processes, and improve success for the way populations are managed in the future.

Animal Health Component

(N/A)

Research Effort Categories

Basic

(N/A)

Applied

(N/A)

Developmental

(N/A)

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
135	0810	2090	10%
135	0819	1070	10%
135	0819	2090	10%
135	0820	2090	10%
135	0830	2090	10%
135	0860	1070	10%
135	0860	2090	10%
135	0899	2090	10%
136	0860	1070	10%
136	0860	2090	10%

Knowledge Area
135 - Aquatic and Terrestrial Wildlife; 136 - Conservation of Biological Diversity;

Subject Of Investigation
0810 - Finfish; 0819 - Natural fisheries, other; 0860 - Endangered species; 0830 - Wild animals; 0899 - Wildlife and natural fisheries, general/other; 0820 - Wild birds;

Field Of Science
1070 - Ecology; 2090 - Statistics, econometrics, and biometrics;

Keywords

population ecology

population dynamics

ecological theory

statistical ecology

multivariate analysis

capture recapture

sampling design

environmental variation

population management

stock assessment

Goals / Objectives
A major part of the proposed project consists of improving the use of statistical methods for the analysis of population data. Thereafter, the improved methods will be applied to a number of existing ecological datasets. These datasets will be wildlife and fish data available from Texas and other parts of the US including simple time-series data (e.g., transect data and catch per unit effort data) and capture-recapture data. The product of the proposed project is a comprehensive set of guidelines for the analysis of population datasets in both applied and basic ecological research.

Project Methods
The proposed project consists of three parts. The first part is to advance statistical methods for analyzing population time series. The second part is the application of the developed methods to actual data. Finally, the third part is to develop general guidelines for collection and analysis of population data. The advances in statistical methods will be achieved by identifying a series of statistical methods for the analysis of population datasets and determining the best order for applying these methods to the available data. This will be primarily achieved through the application of various existing methods to simulated data sets. I will use simulated data sets for two main reasons. First, for simulated data sets, the underlying process that produced the data is known. This allows me to test statistical methods with data sets of known properties. Second, the data in population time series analysis are often short, making asymptotic properties less useful than they are for longer time series, which are often analyzed based on the asymptotic properties. I will begin my assessment of statistical methods by exploring the use of multivariate methods (e.g. principal component factor analysis, independent component factor analysis, maximum autocorrelation factor analysis) to separate the component signals and observational errors from simulated population time series. Multivariate methods are particularly suitable for population data analysis because they take advantage of information in multiple time series rather than that in a single long time series. After researching factor analysis, I will explore the use of state-space methods. State-space methods are statistical approaches for decomposing the variance of time series into various propagating and non-propagating parts, based on the autocorrelation structure of the time series. Finally, in phase one of the project, I will explore the use of capture recapture statistics. Capture-recapture methods require capture/sighting history of individually marked/identified organisms. The technique has proven to be very effective in conservation biology. Once I have identified appropriate statistical methods, I will apply them to actual data. This part of the proposed study will be achieved in two stages. In the first stage, the method will be applied to data from well-studied populations. The idea is that the new method will be used to confirm the known relationships between environmental conditions and population processes, thereby validating the new technique. An obvious first step will be to apply the method to some data available from Texas Park and Wildlife. Then, the research will be expanded to include fish and wildlife data that are have been less studied. Finally, I will produce a comprehensive set of guidelines for the analysis of population data and the type of data to be collected. I anticipate that these guidelines will be structured so that they point to the best statistical approaches for the types of organisms and data (e.g. short-lived vs. long-lived and breeding survey vs. catch data) and the best type of data to be collected for the type of organisms and ecological questions at hand.

Progress 02/01/10 to 01/31/15

Outputs
Target Audience:Conservation Agencies, Fisheries Management Agencies Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?During the study, four graduate students were trained. Two of them graduated: one with M.S. and the other with PhD. Two undergraduate students were also trained. Both of them continue their education as a graduate student. How have the results been disseminated to communities of interest?Research was communicated with scientists at National Marine Fisheries Service in Galveston. What do you plan to do during the next reporting period to accomplish the goals?This is the final report. However, I will continue to publish papers in peer-reviewed journals summarizing the finding.

Impacts
What was accomplished under these goals? First, new approach to infer bottom-up and top-down control in a community from multivariate time series was developed. This approach uses a co-integration method, which is common in economietrics but has not been used in population data analysis. This method was applied to catich-per-unit-effort fish survey data. Second, another approach to identify interactions from time series data, called partial least squares regressions, was applied to environmental data and shrimp survey data to find the effects of environmental conditions on shrimp recruitment. Third, common factor analyses were applied to fishery time series data to demonstrate their use in finding interactions between populations and environment. Finally, several different time series analylses were compared by applying them to simulated data sets. A manusript summarizing the results have been submitted to a peer-reviewed journal. This paper will be a useful guideline for selecting appropriate statistical analysis.

Publications

Progress 10/01/14 to 01/31/15

Outputs
Target Audience:Conservation Agencies, Fisheries Management Agencies Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?Two graduate students and one undergraduate student was involved in the analysis and writing scientific papers. Each of the two graduate students wrote a manuscript, and both of the manuscripts are currently under review by peer-reviewed journals. How have the results been disseminated to communities of interest?Research was communicated with scientists at National Marine Fisheries Service in Galveston. What do you plan to do during the next reporting period to accomplish the goals?This is the last reporting period.

Impacts
What was accomplished under these goals? A major part of the proposed project consists of improving the use of statistical methods for the analysis of population data. Thereafter, the improved methods will be applied to a number of existing ecological datasets. These datasets will be wildlife and fish data available from Texas and other parts of the US including simple time-series data (e.g., transect data and catch per unit effort data) and capture-recapture data. The product of the proposed project is a comprehensive set of guidelines for the analysis of population datasets in both applied and basic ecological research.

Publications

• Type: Journal Articles Status: Published Year Published: 2014 Citation: Fujiwara. M., J.E. Cohen (2014) Mean and variance of population density and temporal Taylors law in stochastic stage-structured density-dependent models of exploited fish populations. Theoretical Ecology

Progress 10/01/13 to 09/30/14

Outputs
Target Audience: Conservation Agencies, Fisheries Management Agencies Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Two graduate students and one undergraduate student was involved in the analysis andwriting scientific papers. Each of the two graduate studentswrote a manuscript, and both of the manuscriptsare currently under review by peer-reviewed journals. How have the results been disseminated to communities of interest? Research was communicated with scientists at National Marine Fisheries Service in Galveston. What do you plan to do during the next reporting period to accomplish the goals? A theoretical study to indentify expected mean and variance of popualtion density will be conducted. This information will be incorporated into the future population data analysis. Another study to try to indentify the potential effect of density and individual size on survival and growth of organisms from time series data will be conducted. Finally, a document summarizing these and other techniques will be written.

Impacts
What was accomplished under these goals? Accomplishements were achieved intwo areas this year. First, new approach to infer bottom-up and top-down control in a community from multivariate time series was developed. This approach uses a co-integration method, which is common in economietrics but has not been used in population data analysis. This method was applied to catich-per-unit-effort fish survey data. Second, another approach to identify interactions from time series data, called partial least squares regressions, was applied to environmental data and shrimp survey data to find the effects of environmental conditionson shrimp recruitment.

Publications

• Type: Journal Articles Status: Published Year Published: 2014 Citation: Baker, R., Fujiwara, M., and Minello, T.J. (2014) Juvenile growth and mortality effects on white shrimp Litopenaeus setiferus population dynamics in the northern Gulf of Mexico. Fisheries Research. 155:74-82.
• Type: Journal Articles Status: Submitted Year Published: 2015 Citation: Zhou, C., Fujiwara, M., Grant, W.E. (2015) Identifying bottom-up and top-down control in marine community dynamics: A new method for analyzing non-stationary, multivariate time series data. Canadian Journal of Fisheries and Aquatic Sciences.
• Type: Journal Articles Status: Submitted Year Published: 2015 Citation: Piper, C.B., Fujiwara, M., Hart, R.A., Nance, J.M. (2015) Effects of tidal height and river discharge on brown Farfatepenaeus aztecus and white Litopenaeus setiferus shrimp recruitment. Transactions of American Fisheries Society.

Progress 01/01/13 to 09/30/13

Outputs
Target Audience: Conservation Agencies, Fisheries Management Agencies Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? During this reporting period, this project was primarily conducted by undergraduate and PhD graduate students. The application of partial least squares regression analysis was conducted by Cyrenea Millberry (senior undergraduate student), and the application of the co-integration method was primarily conducted by Can Zhou (PhD graduate student) under my supervision. How have the results been disseminated to communities of interest? Some of the results were presented at U.S. National Marine Fisheries Service in Galveston, Texas. What do you plan to do during the next reporting period to accomplish the goals? During the next reporting period, I expect to publish two papers based on the results obtained in 2013. In addition, I plan to work on the review paper on time series data analysis outlining the protocol.

Impacts
What was accomplished under these goals? During the reporting period, I have applied various multivariate statistical approaches to actual population data with two general objectives: (1) to demonstrate the use of the techniques and (2) to work toward general guideline for the analyses of population time-series data. My focus was on the analysis of the data from the Southeast Area Monitoring and Assessment Program (SEAMAP). These are fishery-independent catch per unit effort (CPUE) data. First study was to associate shrimp (white shrimp Litopenaeus setiferus and brown shrimp Farfantepenaeus aztecus) CPUE to environmental variables (stream discharge into get Gulf of Mexico and tidal flooding data) using the partial least squared regression analysis. I found that both tide and stream discharge are significantly associated the CPUE’s. This study at a regional scale supports the previous finding by others at local scales. The second study was to associate CPUE of four finfish species (Atlantic croaker Micropogonias undulatus, silver seatrout Cynoscion nothus, spot croaker Leiostomus xanthurus, and sand seatrout Cynoscion arenarius) and brown shrimp using the co-integration method. Co-integration method is another multivariate statistical method, and it allows the analysis of non-stationary time series. In addition, the method allows us to determine the direction of the association (i.e. cause-and-effect relationship). The results suggest there are significant bottom-up effects between shrimp as prey and some of the finfish species as predators.

Publications

• Type: Journal Articles Status: Published Year Published: 2013 Citation: Zhou, C., Fujiwara, M., Grant, W.E. (2013) Dynamics of a predatorprey interaction with seasonal reproductionand continuous predation. Ecological Modelling 268: 25-36
• Type: Journal Articles Status: Published Year Published: 2013 Citation: Fujiwara, M., Zhou, C. (2013) Population dynamics of stage-structured sequential hermaphrodites. Canadian Journal of Fisheries and Aquatic Sciences 70: 1296-1305
• Type: Theses/Dissertations Status: Other Year Published: 2013 Citation: Millberry, C. (2013) Association between Shrimp Catch per Unit Effort and Environmental Variables in the Gulf of Mexico, Undergraduate Research Scholars Thesis, Texas A&M University, College Station, TX
• Type: Conference Papers and Presentations Status: Published Year Published: 2013 Citation: Fujiwara, M. (2013) Size- and Stage-structured Population Model to Assess the Growth-mediated Effect of Environmental Conditions on Population Dynamics, SIAM Conference on Application of Dynamical Systems, May, 2013, Snowbird, UT

Progress 01/01/12 to 12/31/12

Outputs
OUTPUTS: During the reporting period (January 1 to December 31, 2012), I have applied various multivariate statistical approaches to actual population data with two general objectives: (1) to demonstrate the use of the techniques and (2) to work toward general guideline for the analyses of population time-series data. Multivariate statistical technique and factor analyses were applied to the data from the Southeast Area Monitoring and Assessment Program (SEAMAP). These are fishery independent data. In particular, the catch per unit effort (CPUE) of white shrimp Litopenaeus setiferus and brown shrimp Farfantepenaeus aztecus from SEAMAP have been investigated using partial least squares regression analysis and maximum autocorrelation factor analysis. The main objective of the study is to try to identify any association between the CPUE data and environmental variables, such as stream discharge into the Gulf of Mexico and tidal flooding data. The results were directly communicated with researchers at US National Marine Fisheries Service, which oversee the management of the shrimp stocks. PARTICIPANTS: Cyrenea Millberry (undergraduate student) has participated in the project. This is a part of undergraduate research experience. Can Zhou (PhD Graduate Student) also participated in the project. This is a part of his work toward obtaining a PhD degree. TARGET AUDIENCES: Conservation Agencies, Fisheries Management Agencies, Wildlife Management Agencies PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.

Impacts
Standard methods for analyzing population time series include simple correlation analysis and generalized linear models. However, these techniques often lead to spurious results. There is urgent need for improving statistical analysis for population time series data, which are arguably the most common population data available. Many management decisions are also based on such data. My approach, which treats the population data as multivariate time series and applies multivariate technique, will be significant improvement in the analysis method. Once the proposed guideline is completed, it is expected to have a large impact on population time series analysis for scientific as well as management-related studies. The results from the analysis of shrimp data also have immediate impact. Shrimp stocks are economically the second most important natural "fish" stock in the US. Understanding how environmental conditions affect the CPUE will lead to better management of the stocks. This is especially very important because of current plans for restoration/development of coastal areas in the Gulf of Mexico.

Publications

• Fujiwara, M. (2012) Demographic diversity and sustainable fisheries. PLOS One 7(5): e34556 (DOI:10.1371/journal.pone.0034556)
• Anderson, K.E., Fujiwara, M., Rothstein, S.I. (2012) Demography and dispersal of juvenile and adult Brown-headed cowbirds (Molothrus ater) in the eastern Sierra Nevada, California, estimated using multistate models. The Auk 129 (2):307-318 (DOI:10.1525/auk.2012.11164)
• Gil-Weir, K.C., Grant, W.E., Slack, R.D., Wang, H., and Fujiwara, M. (2012) Demography and population trends of Whooping Cranes. Journal of Field Ornithology 83(1): 1-10 (DOI:10.1111/j.1557-9263.2011.00349.x)

Progress 01/01/11 to 12/31/11

Outputs
OUTPUTS: During the reporting period (January 1 to December 31, 2011), I have applied various multivariate statistical approaches to actual population data with two general objectives: (1) to demonstrate the use of the techniques and (2) to work toward general guideline for the analyses of population time-series data. Multivariate statistical technique and factor analyses were applied to spring breeding waterfowl survey data compiled by US Fish and Wildlife Service, and the population data were compared with weather and climate data. In particular, a singular value decomposition method was applied to fill missing data, factor analysis was applied to the breeding bird survey data to extract underlying component signals, and partial least square regression method was applied to all of the data to find potential association between the population and environmental data. Results from this work are currently being written. Factor analysis was also applied to spawning abundance data of Chinook salmon (Oncorhynchus tshawytscha) from tributaries within the Klamath River basin, California, and extracted component signals were compared with stream flow data, which is previously shown to associate with the prevalence of disease-causing agent. The results from this study were previously communicated with stake holders at the Klamath Basin Science Conference in 2010 as a presentation, and its abstract has been published this year (see Publications). The results were directly communicated with researchers at US National Marine Fisheries Service, which oversee the management of the salmon population. PARTICIPANTS: Mengmeng Sun, MS Graduate Student, Wildlife and Fisheries Sciences, Texas A&M University; Can Zhou, PhD Graduate Student, WIldlife and Fisheries Sciences, Texas A&M University TARGET AUDIENCES: Target audiences include federal, state, and private agencies in conservation biology, fisheries management, wildlife management. PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.

Impacts
Standard methods for analyzing population time series include simple correlation analysis and generalized linear models. However, these techniques often lead to spurious results. There is urgent need for improving statistical analysis for population time series data, which are arguably the most common population data available. Many management decisions are also based on such data. My approach, which treats the population data as multivariate time series and applies multivariate technique, will be significant improvement in the analysis method. Once the proposed guideline is completed, it is expected to have a large impact on population time series analysis for scientific as well as management-related studies. The results from the analysis of Chinook salmon data also have immediate impact. The recently discovered high prevalence of Chinook salmon disease within the Klamath Basin is a serious concern. However, general effects of the disease on overall population dynamics have been unknown. My study demonstrated that it will affect spawning abundance and is something that needs to be considered in making management decisions.

Publications

• Fujiwara, M., Pfeiffer, G., Boggess, M., Day, S., Walton, J. (2011). Coexistence of competing stage-structured populations. Scientific Reports, 1, 107 (DOI:10.1038/srep00107).
• Fujiwara, M., Mohr, M.S., Greenberg, A., Foott, J.S., Bartholomew, J.L. (2011). Effects of Ceratomyxosis on population dynamics of Klamath fall-run Chinook salmon. Transactions of American Fisheries Society, 140:1380-1391 (DOI: 10.1080/00028487.2011.621811).

Progress 01/01/10 to 12/31/10

Outputs
OUTPUTS: Activities: During the reporting period (Jan 1, 2010 – Dec 31, 2010), Fujiwara investigated two population data sets applying several multivariate techniques. The first data set is spawning abundance of Pacific salmon. The analysis of the data was accomplished by applying a transformation technique to emphasize relative success of populations between tributaries and applying a factor analysis. The second data set is breeding waterfowl counts from various locations in the US and Canada. The goal of the project is to improve/develop a method for associating the waterfowl population data and environmental variables such as temperature and precipitation. To this extent, waterfowl data were obtained from US Fish and Wildlife Service and environmental variables were obtained from US Weather Service. Fujiwara is currently developing a detailed plan for the analysis. Fujiwara taught an undergraduate course in Marine Fisheries (WFSC 425, 18 Students) and graduate course in Dynamics of Population (WFSC 624, 8 Students). Fujiwara also mentored two undergraduate students in 2010 to conduct undergraduate research. Fujiwara also prepared for distance education to be started in 2011. Fujiwara participate in Undergraduate Mathematics and Biology program to provide quantitative research experience to undergraduate students. Products: Fujiwara networked with scientists at US National Marine Fisheries Service in Galveston. This interaction is expected to develop into future collaborative work. Fujiwara developed and undergraduate curriculum in Quantitative Ecology Emphasis to provide opportunity to students in Wildlife and Fisheries Students to learn more quantitative methods. PARTICIPANTS: Mengmeng Sun, MS graduate student. TARGET AUDIENCES: The target audience of Marine Fisheries (WFSC 425) is undergraduate students in fisheries sciences. The target audence of Dynamics of Population (WFSC 624) is graduate students in wilelife and fisheries. PROJECT MODIFICATIONS: N/A

Impacts
Using a multivariate time series data, Fujiwara discovered a potential association between disease-causing agent and survival of juvenile salmon. This is expected to lead to change in the way salmon population is managed and environmental impact assessment. The statistical analysis used in the analysis of salmon data is also expected to change the way population time series are analyzed in the future.

Publications

• Kendall, B.E., Fox, G.A., Fujiwara, M. Nogeire, T. (2011) Demographic heterogeneity, cohort selection and population growth. Ecology. (pending)
• Fujiwara, M., Mohr, M.S., Greenberg, A., Foott, J.S., Bartholomew, J.L. (2011) Effects of Ceratomyxosis on Population Dynamics of Pacific Salmon. Transaction of American Fisheries Society. (provisionally accepted)