DSpace Community:http://hdl.handle.net/10023/6252015-05-05T16:19:22Z2015-05-05T16:19:22ZWorkshop on new developments in cetacean survey methodsBorchers, David LouisThomas, LenBuckland, Stephen TerrenceSkaug, HansBarlow, Jayhttp://hdl.handle.net/10023/32162014-06-23T12:31:02Z2011-01-01T00:00:00ZAbstract: This report contains the slides from a workshop on New Developments in Cetacean Survey Methods held on 27th November 2011 at the 19th Biennial Conference on the Biology of Marine Mammals, Tampa, Florida. Review talks were given on Passive Acoustic Density Estimation (Len Thomas); Dealing with g(0)<1: Perception Bias (Stephen Buckland); Dealing with g(0)<1: Availability Bias (Hans Skaug); Dealing with Measurement Error (David Borchers); and Density Surface Modelling (Jay Barlow). The sessions were followed by a discussion, and this is summarized at the end of the report.2011-01-01T00:00:00ZBorchers, David LouisThomas, LenBuckland, Stephen TerrenceSkaug, HansBarlow, JayThis report contains the slides from a workshop on New Developments in Cetacean Survey Methods held on 27th November 2011 at the 19th Biennial Conference on the Biology of Marine Mammals, Tampa, Florida. Review talks were given on Passive Acoustic Density Estimation (Len Thomas); Dealing with g(0)<1: Perception Bias (Stephen Buckland); Dealing with g(0)<1: Availability Bias (Hans Skaug); Dealing with Measurement Error (David Borchers); and Density Surface Modelling (Jay Barlow). The sessions were followed by a discussion, and this is summarized at the end of the report.A critical review of the literature on population modellingCabrelli, AbigailHarwood, JohnMatthiopoulos, JasonNew, Leslie FrancesThomas, Lenhttp://hdl.handle.net/10023/22412014-06-26T14:01:02Z2009-01-01T00:00:00ZAbstract: The 2005 report of the National Research Council’s ‘Committee on Characterizing Biologically Significant Marine Mammal Behavior’ proposed a framework, which they called PCAD - Population Consequences of Acoustic Disturbance, that uses a series of transfer functions to link behavioural responses to sound with life functions, vital rates, and population change. The Committee suggested that the best understood transfer functions are those linking vital rates to population change. One of the main aims of this report is to document that understanding. However, we also show how the existing frameworks for modelling the dynamics of marine mammal populations can be extended to include the effects of behavioural responses on vital rates. In Chapter 1 we introduce the central concept of the rate of increase (lambda) of a population, which we believe is the most useful measure of the effects of behavioural responses on the dynamics of a population. If the value of lambda exceeds one, then thepopulation will increase over time; if it is less than one it will decrease. We show how changes in lambda provide a measure of the impact of human activities (such as exploitation, conservation, or disturbance) on a population. We also introduce structured population models, which take account of the fact that all individuals in a population are not identical, and show how the dynamics of different parts of a population can be modelled using a population projection matrix. The mathematical properties of this projection matrix can be used to determine the sensitivity of lambda to small changes in vital rates. Finally, we provide a very brief introduction to the concept of stochasticity, and the use of lambda to predict when (and if) a population might be driven to extinction. Chapter 2 describes how lambda also provides a measure of the Darwinian fitness of the individual members of a population. An individual’s fitness, the contribution it will make to future generations, depends to a large extent on its body condition and on the risks of mortality to which it is exposed. Both of these could be affected by behaviour responses to sound. We also explain current theories about the relationship between an individual’s feeding behaviour and the abundance and distribution of prey, and how this can affect body condition. Chapter 3 provides a more detailed description of how elasticity analysis can be used to investigate the impact of changes in vital rates on lambda . Elasticity analysis is a useful tool for detecting which vital rates are most important in determining the dynamics of a population. However, its value is limited because it does not take account of random variations (stochasticity) and, in theory, it can only predict the effect of small changes in vital rates. Chapter 4 describes the fundamental concept of density dependence: the way in which vital rates change with population size or the availability of resources, such as prey. Not only is density dependence an essential prerequisite for population stability and sustainable use, but the form it takes will also determine how a population responds to behavioural changes. This is because behaviour, and particularly the effect of behavioural change on body condition, plays a central role in many of the mechanistic models of density dependence. Chapters 5 and 6 explore the way in which additional complexities, such as social structure and the way in which populations are distributed in space, can affect the dynamics of populations. Models that account for these complexities behave in a much less predictable way than the relatively simple structured models that form the core of Chapters 1-4. So far, the models of population dynamics that we have reviewed have been deterministic. That is, they have assumed that the only way in which vital rates can vary is in response to a change in abundance, via density dependent mechanisms. In Chapters 7 and 8 we investigate the effect of random variation (stochasticity) on population dynamics. We distinguish the effects of demographic stochasticity, chance variations in the number of animals that die or give birth in a time interval that occur even if vital rates do not vary over time, and environmental stochasticity, which is the result of variations in vital rates across years. Variation in abundance may also occur as a result of environmental change and changes in the ecological community of which a population is a part. The effect of all these sources of variation is to reduce the realised growth rate of a population, and therefore its risk of extinction. In Chapter 9 we consider how the basic population modelling framework described in Chapters 1-8 might be extended to take account of the life functions identified by the NRC Committee. We suggest that these life functions are useful for defining the context in which behavioural responses might affect vital rates, but that they do not need to be modelled explicitly. Removing vital functions from the PCAD framework results in a much simpler structure, which is compatible with existing population modelling frameworks. However, these will have to be extended to allow population states, like body condition, that vary continuously to be modelled. Chapter 10 describes how changes in lambda can be detected. The simple analytical frameworks that are available for this are all vulnerable to the effects of variability that we introduced in Chapter 7. However, there is a framework (state-space and hidden Markov process modelling) that can account for the effects of this variability, and we recommend its use for detecting trends. The additional benefit of this approach is that its use results in a detailed model of the dynamics of the population that is under investigation. Chapter 11 reviews the different model structures that can be used to describe the dynamics of a population, and explains when different forms of population models (e.g. discrete vs. continuous time, deterministic vs. stochastic) are most appropriate. We also discuss how these different frameworks can be extended to account for continuous population states, as recommended in Chapter 8. The final focus is on how state-space models can be fitted to time series of abundance estimates and information on vital rates. Chapter 12 looks at the relevance of the different modelling approaches described in the previous chapters for analysing the potential effects of behavioural responses to sound on population dynamics, particularly the kinds of sounds that may be generated by the oil and gas industry. We conclude that lambda , the population rate of increase, and its variation provides a useful measure of these effects. We also believe that the models used for this purpose will certainly have to account for the effects of variability and density dependence. They will probably also have to account for the effects of social structure and the way in which populations use space. The state-space modelling framework outlined in Chapter 11 can, in principle, be extended to capture all of these features although work on this is still in its infancy.
Description: Final Report to the Joint Industry Project of the International Association of Oil & Gas Producers on contract JIP22 07_202009-01-01T00:00:00ZCabrelli, AbigailHarwood, JohnMatthiopoulos, JasonNew, Leslie FrancesThomas, LenThe 2005 report of the National Research Council’s ‘Committee on Characterizing Biologically Significant Marine Mammal Behavior’ proposed a framework, which they called PCAD - Population Consequences of Acoustic Disturbance, that uses a series of transfer functions to link behavioural responses to sound with life functions, vital rates, and population change. The Committee suggested that the best understood transfer functions are those linking vital rates to population change. One of the main aims of this report is to document that understanding. However, we also show how the existing frameworks for modelling the dynamics of marine mammal populations can be extended to include the effects of behavioural responses on vital rates. In Chapter 1 we introduce the central concept of the rate of increase (lambda) of a population, which we believe is the most useful measure of the effects of behavioural responses on the dynamics of a population. If the value of lambda exceeds one, then thepopulation will increase over time; if it is less than one it will decrease. We show how changes in lambda provide a measure of the impact of human activities (such as exploitation, conservation, or disturbance) on a population. We also introduce structured population models, which take account of the fact that all individuals in a population are not identical, and show how the dynamics of different parts of a population can be modelled using a population projection matrix. The mathematical properties of this projection matrix can be used to determine the sensitivity of lambda to small changes in vital rates. Finally, we provide a very brief introduction to the concept of stochasticity, and the use of lambda to predict when (and if) a population might be driven to extinction. Chapter 2 describes how lambda also provides a measure of the Darwinian fitness of the individual members of a population. An individual’s fitness, the contribution it will make to future generations, depends to a large extent on its body condition and on the risks of mortality to which it is exposed. Both of these could be affected by behaviour responses to sound. We also explain current theories about the relationship between an individual’s feeding behaviour and the abundance and distribution of prey, and how this can affect body condition. Chapter 3 provides a more detailed description of how elasticity analysis can be used to investigate the impact of changes in vital rates on lambda . Elasticity analysis is a useful tool for detecting which vital rates are most important in determining the dynamics of a population. However, its value is limited because it does not take account of random variations (stochasticity) and, in theory, it can only predict the effect of small changes in vital rates. Chapter 4 describes the fundamental concept of density dependence: the way in which vital rates change with population size or the availability of resources, such as prey. Not only is density dependence an essential prerequisite for population stability and sustainable use, but the form it takes will also determine how a population responds to behavioural changes. This is because behaviour, and particularly the effect of behavioural change on body condition, plays a central role in many of the mechanistic models of density dependence. Chapters 5 and 6 explore the way in which additional complexities, such as social structure and the way in which populations are distributed in space, can affect the dynamics of populations. Models that account for these complexities behave in a much less predictable way than the relatively simple structured models that form the core of Chapters 1-4. So far, the models of population dynamics that we have reviewed have been deterministic. That is, they have assumed that the only way in which vital rates can vary is in response to a change in abundance, via density dependent mechanisms. In Chapters 7 and 8 we investigate the effect of random variation (stochasticity) on population dynamics. We distinguish the effects of demographic stochasticity, chance variations in the number of animals that die or give birth in a time interval that occur even if vital rates do not vary over time, and environmental stochasticity, which is the result of variations in vital rates across years. Variation in abundance may also occur as a result of environmental change and changes in the ecological community of which a population is a part. The effect of all these sources of variation is to reduce the realised growth rate of a population, and therefore its risk of extinction. In Chapter 9 we consider how the basic population modelling framework described in Chapters 1-8 might be extended to take account of the life functions identified by the NRC Committee. We suggest that these life functions are useful for defining the context in which behavioural responses might affect vital rates, but that they do not need to be modelled explicitly. Removing vital functions from the PCAD framework results in a much simpler structure, which is compatible with existing population modelling frameworks. However, these will have to be extended to allow population states, like body condition, that vary continuously to be modelled. Chapter 10 describes how changes in lambda can be detected. The simple analytical frameworks that are available for this are all vulnerable to the effects of variability that we introduced in Chapter 7. However, there is a framework (state-space and hidden Markov process modelling) that can account for the effects of this variability, and we recommend its use for detecting trends. The additional benefit of this approach is that its use results in a detailed model of the dynamics of the population that is under investigation. Chapter 11 reviews the different model structures that can be used to describe the dynamics of a population, and explains when different forms of population models (e.g. discrete vs. continuous time, deterministic vs. stochastic) are most appropriate. We also discuss how these different frameworks can be extended to account for continuous population states, as recommended in Chapter 8. The final focus is on how state-space models can be fitted to time series of abundance estimates and information on vital rates. Chapter 12 looks at the relevance of the different modelling approaches described in the previous chapters for analysing the potential effects of behavioural responses to sound on population dynamics, particularly the kinds of sounds that may be generated by the oil and gas industry. We conclude that lambda , the population rate of increase, and its variation provides a useful measure of these effects. We also believe that the models used for this purpose will certainly have to account for the effects of variability and density dependence. They will probably also have to account for the effects of social structure and the way in which populations use space. The state-space modelling framework outlined in Chapter 11 can, in principle, be extended to capture all of these features although work on this is still in its infancy.An update to the methods in Endangered Species Research 2011 paper "Estimating North Pacific right whale Eubalaena japonica density using passive acoustic cue counting"Marques, Tiago A.Munger, LisaThomas, LenWiggins, SeanHildebrand, Johnhttp://hdl.handle.net/10023/21582014-06-16T23:02:12Z2012-01-01T00:00:00Z2012-01-01T00:00:00ZMarques, Tiago A.Munger, LisaThomas, LenWiggins, SeanHildebrand, JohnComparing pre- and post-construction distributions of long-tailed ducks Clangula hyemalis in and around the Nysted offshore wind farm, Denmark : a quasi-designed experiment accounting for imperfect detection, local surface features and autocorrelationPetersen, Ib KragMacKenzie, Monique LeaRexstad, EricWisz, Mary S.Fox, Anthony D.http://hdl.handle.net/10023/20082014-06-23T15:01:01Z2011-01-01T00:00:00ZAbstract: We report a novel technique to model abundance patterns of wintering seaducks in relation to the construction of an offshore wind farm (OWF) based on seven years of aerial survey transect data. Distance sampling was used to estimate seaduck densities adjusted for covariates affecting detection probabilities. A generalized additive model (GAM) generated seaduck densities in sampling units in relation to spatially explicit covariates, using bootstrapping to account for uncertainties in both processes. Generalized estimating equations generated precision measures for the GAM robust to spatial and temporal autocorrelation. Comparison of pre- and post-construction model generated surfaces showed significant reductions in long-tailed duck numbers only within the OWF (despite the fact that the model was uninformed about the OWF location), although the absolute numbers involved were trivial in a flyway population context. This method provides quantification of distributional effects on organisms over a gradient in space and time that offers an alternative to Before-After/Control-Impact designs in environmental impact assessment.2011-01-01T00:00:00ZPetersen, Ib KragMacKenzie, Monique LeaRexstad, EricWisz, Mary S.Fox, Anthony D.We report a novel technique to model abundance patterns of wintering seaducks in relation to the construction of an offshore wind farm (OWF) based on seven years of aerial survey transect data. Distance sampling was used to estimate seaduck densities adjusted for covariates affecting detection probabilities. A generalized additive model (GAM) generated seaduck densities in sampling units in relation to spatially explicit covariates, using bootstrapping to account for uncertainties in both processes. Generalized estimating equations generated precision measures for the GAM robust to spatial and temporal autocorrelation. Comparison of pre- and post-construction model generated surfaces showed significant reductions in long-tailed duck numbers only within the OWF (despite the fact that the model was uninformed about the OWF location), although the absolute numbers involved were trivial in a flyway population context. This method provides quantification of distributional effects on organisms over a gradient in space and time that offers an alternative to Before-After/Control-Impact designs in environmental impact assessment.Density estimation implications of increasing ambient noise on beaked whale click detection and classificationMarques, Tiago Andre Lamas OliveiraWard, JessicaJarvis, SusanMoretti, DavidMorrissey, RonaldDiMarzio, NancyThomas, Lenhttp://hdl.handle.net/10023/16522014-05-21T13:31:01Z2010-01-01T00:00:00ZAbstract: Acoustic based density estimates are being increasingly used. Usually density estimation methods require one to evaluate the eﬀective survey area of the acoustic sensors, or equivalently estimate the mean detection probability of detecting the animals or cues of interest. This is often done based on an estimated detection function, the probability of detecting an object of interest as a function of covariates, usually distance and additional covariates. If the actual survey data and the data used to estimate a detection function are not collected simultaneously, as in Marques et al. (2009), the estimated detection function might not correspond to the detection process that generated the survey data. This would lead to biaseddensity estimates. Here we evaluate the inﬂuence of ambient noise in the detection and classiﬁcation of beaked whale clicks at the Atlantic Undersea Test and Evaluation Center (AUTEC) hydrophones, to assess if the density estimates reported in Marques et al. (2009) might have been biased. To do so we contaminated a data set with increasing levels of ambient noise, and then estimated the detection function accounting for the noise level as an additional covariate. The results obtained suggest that for the particular results obtained at AUTEC’s deep water hydrophones the inﬂuence of ambient noise on the beaked whale’s click detection probability might have been minor, and hence unlikely to have had an impact on density estimates. However, we do not exclude the possibility that the results could be diﬀerent under other scenarios.2010-01-01T00:00:00ZMarques, Tiago Andre Lamas OliveiraWard, JessicaJarvis, SusanMoretti, DavidMorrissey, RonaldDiMarzio, NancyThomas, LenAcoustic based density estimates are being increasingly used. Usually density estimation methods require one to evaluate the eﬀective survey area of the acoustic sensors, or equivalently estimate the mean detection probability of detecting the animals or cues of interest. This is often done based on an estimated detection function, the probability of detecting an object of interest as a function of covariates, usually distance and additional covariates. If the actual survey data and the data used to estimate a detection function are not collected simultaneously, as in Marques et al. (2009), the estimated detection function might not correspond to the detection process that generated the survey data. This would lead to biaseddensity estimates. Here we evaluate the inﬂuence of ambient noise in the detection and classiﬁcation of beaked whale clicks at the Atlantic Undersea Test and Evaluation Center (AUTEC) hydrophones, to assess if the density estimates reported in Marques et al. (2009) might have been biased. To do so we contaminated a data set with increasing levels of ambient noise, and then estimated the detection function accounting for the noise level as an additional covariate. The results obtained suggest that for the particular results obtained at AUTEC’s deep water hydrophones the inﬂuence of ambient noise on the beaked whale’s click detection probability might have been minor, and hence unlikely to have had an impact on density estimates. However, we do not exclude the possibility that the results could be diﬀerent under other scenarios.Comparison of aerial survey methods for estimating abundance of common scotersRexstad, EricBuckland, Stephen T.http://hdl.handle.net/10023/7842012-12-12T11:04:28Z2009-01-01T00:00:00ZAbstract: During the month of March, four survey methods were applied to the SPA at Camarthen Bay. WWT staff carried out visual aerial surveys using distance sampling methodology (Camphuysen et al. 2004). Visual shore-based counts were also conducted. Distance measures were not consistently taken by these observers, nor was survey effort equal among the four surveys. Because they are intended to be complete counts without replication within a day, it is not possible to estimate precision of these counts, or assess bias, making comparison with other survey results difficult. Digital still data were collected and processed by APEM Ltd. Digital video imagery were captured and processed by HiDef. This report revision includes 29 March survey data from HiDef not available at the time of the release of our 17 July report.2009-01-01T00:00:00ZRexstad, EricBuckland, Stephen T.During the month of March, four survey methods were applied to the SPA at Camarthen Bay. WWT staff carried out visual aerial surveys using distance sampling methodology (Camphuysen et al. 2004). Visual shore-based counts were also conducted. Distance measures were not consistently taken by these observers, nor was survey effort equal among the four surveys. Because they are intended to be complete counts without replication within a day, it is not possible to estimate precision of these counts, or assess bias, making comparison with other survey results difficult. Digital still data were collected and processed by APEM Ltd. Digital video imagery were captured and processed by HiDef. This report revision includes 29 March survey data from HiDef not available at the time of the release of our 17 July report.Estimating the distribution of demersal fishing effort from VMS data using hidden Markov models.Borchers, David L.Reid, David G.http://hdl.handle.net/10023/6362011-02-22T14:32:03Z2008-01-01T00:00:00ZDescription: Previously in the University eprints HAIRST pilot service at http://eprints.st-andrews.ac.uk/archive/00000461/2008-01-01T00:00:00ZBorchers, David L.Reid, David G.Incorporating Model Uncertainty into the Sequential Importance Sampling Framework using a Model Averaging Approach, or Trans-Dimensional Sequential Importance Sampling.Lynam, ChristopherKing, Ruth, 1977-Thomas, LenBuckland, Stephen T.http://hdl.handle.net/10023/6352010-12-14T09:07:09Z2007-01-01T00:00:00ZAbstract: A sequential Bayesian Monte Carlo approach is proposed in which model space can be explored during the Sequential Importance Sampling (SIS, a.k.a. Particle Filtering) fitting process. The algorithm allows model space to be explored while filtering forwards through time and takes a similar approach to Reversible Jump Markov Chain Monte Carlo (RJMCMC) strategies, whereby parameters jump into and out of the model structure. Possible efficiency gains of the new Trans-Dimensional SIS routine are discussed and the approach is considered most beneficial when the exploration of large model space in the SIS framework is desired.
Description: Previously in the University eprints HAIRST pilot service at http://eprints.st-andrews.ac.uk/archive/00000463/2007-01-01T00:00:00ZLynam, ChristopherKing, Ruth, 1977-Thomas, LenBuckland, Stephen T.A sequential Bayesian Monte Carlo approach is proposed in which model space can be explored during the Sequential Importance Sampling (SIS, a.k.a. Particle Filtering) fitting process. The algorithm allows model space to be explored while filtering forwards through time and takes a similar approach to Reversible Jump Markov Chain Monte Carlo (RJMCMC) strategies, whereby parameters jump into and out of the model structure. Possible efficiency gains of the new Trans-Dimensional SIS routine are discussed and the approach is considered most beneficial when the exploration of large model space in the SIS framework is desired.Accommodating availability bias on line transect surveys using hidden Markov models.Borchers, David L.Samara, Filipa I. P.http://hdl.handle.net/10023/6332009-03-24T12:57:39Z2007-01-01T00:00:00ZAbstract: Maximum likelihood methods are developed which accommodate intermittent animal availability of animals on line transect surveys. Existing 'availability bias' correction methods are shown to be inadequate in general. The new method is applied to an aerial survey of whales, using a hidden Markov model to characterise the availability process.
Description: Previously in the University eprints HAIRST pilot service at http://eprints.st-andrews.ac.uk/archive/00000458/2007-01-01T00:00:00ZBorchers, David L.Samara, Filipa I. P.Maximum likelihood methods are developed which accommodate intermittent animal availability of animals on line transect surveys. Existing 'availability bias' correction methods are shown to be inadequate in general. The new method is applied to an aerial survey of whales, using a hidden Markov model to characterise the availability process.Investigation of towed hydrophone monitoring power for harbour porpoise on the SCANS II survey.Borchers, David L.Burt, M. Louise.http://hdl.handle.net/10023/6322009-01-14T15:14:31Z2007-01-01T00:00:00ZAbstract: We investigate the power of harbour porpoise monitoring programmes which use an index of relative abundance to detect change. Power depends on the variability in the constant of proportionality relating the index to absolute abundance, as well as on the variability in the index given this constant. We estimate both from the SCANS II data and from European Seabirds at Sea (ESAS) data. Estimates of the coefficient of variation of the constant of proportionality are large and this results in very low power. Because these estimates may be unrealistically large for well-designed monitoring programs, we feel it is inappropriate to draw strong conclusions about the power of future monitoring programmes based on them.
ESAS surveys are found to be more efficient in terms of effort required to achieve given power, than the SCANS II passive acoustic surveys. However, the comparison may not be a fair one, for the following reason. The estimated CV of the constant of proportionality is obtained from the ratio of the index of density and the corresponding SCANS II absolute density estimate; the ESAS index is likely to be more highly correlated with the SCANS II estimate than the acoustic index, because like the SCANS II estimate, it is based on visual detections. In addition, standardization of the passive acoustic survey methods could yield substantially higher efficiency.
We provide a table giving power as a function of the CV of the constant of proportionality and the CV of the index, given this constant - this can be used to compare methods if reliable estimates of these CVs are available.
Description: Previously in the University eprints HAIRST pilot service at http://eprints.st-andrews.ac.uk/archive/00000457/2007-01-01T00:00:00ZBorchers, David L.Burt, M. Louise.We investigate the power of harbour porpoise monitoring programmes which use an index of relative abundance to detect change. Power depends on the variability in the constant of proportionality relating the index to absolute abundance, as well as on the variability in the index given this constant. We estimate both from the SCANS II data and from European Seabirds at Sea (ESAS) data. Estimates of the coefficient of variation of the constant of proportionality are large and this results in very low power. Because these estimates may be unrealistically large for well-designed monitoring programs, we feel it is inappropriate to draw strong conclusions about the power of future monitoring programmes based on them.
ESAS surveys are found to be more efficient in terms of effort required to achieve given power, than the SCANS II passive acoustic surveys. However, the comparison may not be a fair one, for the following reason. The estimated CV of the constant of proportionality is obtained from the ratio of the index of density and the corresponding SCANS II absolute density estimate; the ESAS index is likely to be more highly correlated with the SCANS II estimate than the acoustic index, because like the SCANS II estimate, it is based on visual detections. In addition, standardization of the passive acoustic survey methods could yield substantially higher efficiency.
We provide a table giving power as a function of the CV of the constant of proportionality and the CV of the index, given this constant - this can be used to compare methods if reliable estimates of these CVs are available.Methods for estimating sperm whale abundance from passive acoustic line transect surveys.Borchers, David L.Brewer, CiaraMatthews, Justinhttp://hdl.handle.net/10023/6312010-11-10T13:35:13Z2007-01-01T00:00:00ZDescription: Previously in the University eprints HAIRST pilot service at http://eprints.st-andrews.ac.uk/archive/00000456/2007-01-01T00:00:00ZBorchers, David L.Brewer, CiaraMatthews, JustinPoint and interval estimates of abundance using multiple covariate distance sampling: an example using great bustards.Rexstad, Erichttp://hdl.handle.net/10023/6292009-11-13T15:33:58Z2007-01-01T00:00:00ZAbstract: Description of computations to produce sex-specific estimates of density from a multiple-covariate distance sampling analysis. Program Distance 5.0 has limited capacity to bootstrap certain types of analytical situations (e.g., cluster size as a covariate). Herein I describe steps and code to perform an analysis of this sort. Possible ways to adapt this code for similar analyses are described.
Description: Previously in the University eprints HAIRST pilot service at http://eprints.st-andrews.ac.uk/archive/00000447/; The pdf file contains the tech report, the ASCII (.R) file contains the accompanying R code.2007-01-01T00:00:00ZRexstad, EricDescription of computations to produce sex-specific estimates of density from a multiple-covariate distance sampling analysis. Program Distance 5.0 has limited capacity to bootstrap certain types of analytical situations (e.g., cluster size as a covariate). Herein I describe steps and code to perform an analysis of this sort. Possible ways to adapt this code for similar analyses are described.Non-uniform coverage estimators for distance sampling.Rexstad, Erichttp://hdl.handle.net/10023/6282009-11-13T15:34:30Z2007-01-01T00:00:00ZAbstract: Allocation of sampling effort in the context of distance sampling is considered.
Specifically, allocation of effort in proportion to portions of the survey region that likely
contain high concentrations of animals are explored. The probability of a portion of the
survey region being included in the sample is proportional to the estimated number of
animals in that portion. These estimated numbers of animals may be derived from a
density surface model. This results in unequal coverage probability, and a Horvitz-
Thompson like estimator can be used to estimate population abundance. The properties
of this estimator is explored here via simulation. The benefits, measured in terms of
increased precision over traditional equal coverage probability estimators, are meagre,
and largely manifested when the underlying population distribution is a smooth gradient.
Description: Previously in the University eprints HAIRST pilot service at http://eprints.st-andrews.ac.uk/archive/00000445/2007-01-01T00:00:00ZRexstad, EricAllocation of sampling effort in the context of distance sampling is considered.
Specifically, allocation of effort in proportion to portions of the survey region that likely
contain high concentrations of animals are explored. The probability of a portion of the
survey region being included in the sample is proportional to the estimated number of
animals in that portion. These estimated numbers of animals may be derived from a
density surface model. This results in unequal coverage probability, and a Horvitz-
Thompson like estimator can be used to estimate population abundance. The properties
of this estimator is explored here via simulation. The benefits, measured in terms of
increased precision over traditional equal coverage probability estimators, are meagre,
and largely manifested when the underlying population distribution is a smooth gradient.