Darren Green's publications

expand all

Refereed journal papers

In press

abstract Soares, S., Murray, A.G., Crumlish, M., Turnbull, J.F. & Green, D.M. (2013). Factors affecting variation in mortality of marine Atlantic salmon (Salmo salar) in Scotland. Diseases of Aquatic Organisms, in press.
Databases of site production have an important role to play in the investigation and understanding of diseases, since they store valuable amounts of disease and management data. Diseases pose an important constraint to economic expansion of aquaculture: they are dependent on the complex interacting factors of pathogen, environment, and host, and the causes of death can be related to nutritional, environmental, and genetic factors of the host or infectious agents. We examined the drivers for mortality from a single site-production database, which represented one third of Scottish farmed salmon production in 2005, to determine whether mortality "benchmarking" data could be generalised across sites and production cycles. We show that farm mortality records play an important role in studying mortality losses and for identification of management problems in production. We found that mortalities varied across the months of the year and with the time of year of initial stocking. Production cycles that started in the third quarter of the year had the highest mortality overall. Furthermore, we found site-to-site variation in mortality that may be caused by either random occurrence of epidemics and environmental events, or other local effects.
abstract Oidtmann, B., Verner-Jeffreys, D., Pond, M., Peeler, E.J., Noguera, P., Bruno, D., LaPatra, S.E., St-Hilaire, S., Schubinger, C., Snekvik, K., Crumlish, M., Green, D.M., Metselaar, M., Rodger, H., Schmidt-Posthaus, H., Galeotti, M., & Feist, S.W. (2013).Differential characterisation of emerging skin diseases of rainbow trout - a standardised approach to capturing disease characteristics and developing case definitions. Journal of Fish Diseases, in press.



abstract Soares, S., Murray, A.G., Crumlish, M., Turnbull, J.F. & Green, D.M. (2012). Evaluating elevated mortality as an indicator of disease presence in the Atlantic salmon industry using the receiver operating characteristic (ROC). Aquaculture, 370-371: 136-143.
Aquaculture faces many threats, including diseases, of which some are notifiable under current UK regulation, e.g. infectious salmon anaemia (ISA) and infectious haematopoietic necrosis (IHN). Abnormal mortality is one possible indicator of the presence of infectious disease on a site that may be used, by the regulator, as a surveillance alert that allows them to identify possible notifiable diseases and to activate measures of control to reduce the risk of spreading those diseases. Therefore, mortality records at the farm level may be a useful indicator for regulatory surveillance purposes in order to identify potential disease outbreaks. In the UK, regulators and producers have discussed abnormal rates of mortality that may be considered as a trigger to notify the official regulator. In our study, the receiver operating characteristic (ROC) approach was used on mortality data from production cycles of a site production database of marine Atlantic salmon belonging to a single company. The usefulness of these data in helping the detection of infectious diseases was determined using measures of sensitivity and specificity. For fish under 750 g, the abnormal rates of mortality did not provide a strong indication of the presence of disease. The area under the curve (0 ≤ AUC ≤ 1) values were generally low with the exception of cardiomyopathy syndrome (CMS) that showed AUC = 0.77 for weekly mortality and AUC = 0.73 for five-week rolling mortality. However, abnormal levels of mortality for fish with weight over 750 g provided a strong indication of the presence of disease with the exception of both suspected and confirmed IPN. The probabilities of triggering official notification were low since mortality events over the percentages proposed happened infrequently. The most efficient trigger will be for weekly mortality (1%) for fish with weight over 750 g since abnormal mortalities in such large fish are more likely to be associated with the presence of disease.
abstract Stirling STORRE Green, D.M., Penman, D.J., Migaud, Bron, J.E., Taggart, J.B. & McAndrew, B.J. (2012). The impact of escaped farmed Atlantic salmon (Salmo salar L.) on catch statistics in Scotland. PLoS One, 7(9): e43560.
In Scotland and elsewhere, there are concerns that escaped farmed Atlantic salmon (Salmo salar L.) may impact on wild salmon stocks. Potential detrimental effects could arise through disease spread, competition, or inter-breeding. We investigated whether there is evidence of a direct effect of recorded salmon escape events on wild stocks in Scotland using anglers' counts of caught salmon (classified as wild or farmed) and sea trout (Salmo trutta L.). This tests specifically whether documented escape events can be associated with reduced or elevated escapes detected in the catch over a five-year time window, after accounting for overall variation between areas and years. Alternate model frameworks were somewhat inconsistent, however no robust association was found between documented escape events and higher proportion of farm-origin salmon in anglers' catch, nor with overall catch size. A weak positive correlation was found between local escapes and subsequent sea trout catch. This is in the opposite direction to what would be expected if salmon escapes negatively affected wild fish numbers. Our approach specifically investigated documented escape events, contrasting with earlier studies examining potentially wider effects of salmon farming on wild catch size. This approach is more conservative, but alleviates some potential sources of confounding, which are always of concern in observational studies. Successful analysis of anglers' reports of escaped farmed salmon requires high data quality, particularly since reports of farmed salmon are a relatively rare event in the Scottish data. Therefore, as part of our analysis, we reviewed studies of potential sensitivity and specificity of determination of farmed origin. Specificity estimates are generally high in the literature, making an analysis of the form we have performed feasible.
abstract Calloway, R.M., Shinn, A.P., Grenfell, S.E., Bron, J.E., Brunell, G., Cook, E.J., Crumlish, M., Culloty, S., Davidson, K., Ellis, R.P., Flynn, K.J., Fox, C., Green, D.M., Hays, G.C., Hughes, A.D., Johnston, E., Lowe, C.D., Lupatsch, I., Malham, S., Mendzil, A.F., Nickell, T., Pickerell, T., Rowley, A.F., Stanley, M.S., Tocher, D.R., Turnbull, J.F., Webb, G., Wootton, E., & Shields, R. (2012). Review of climate change and marine aquaculture in the UK and Ireland. Aquatic Conservation: Marine and Freshwater Ecosystems 22: 389-421.

Marine aquaculture relies on coastal habitats that will be affected by climate change. This review assesses current knowledge of the threats and opportunities of climate change for aquaculture in the UK and Ireland, focusing on the most commonly farmed species, blue mussels (Mytilus edulis) and Atlantic salmon (Salmo salar).

There is sparse evidence to indicate that climate change is affecting aquaculture in the UK and Ireland. Impacts to date have been difficult to discern from natural environmental variability, and the pace of technological development in aquaculture overshadows effects of climatic change. However, this review of broader aquaculture literature and the likely effects of climate change suggests that over the next century, climate change has the potential to directly impact the industry.

Impacts are related to the industry’s dependence on the marine environment for suitable biophysical conditions. For instance, changes in the frequency and strength of storms pose a risk to infrastructure, such as salmon cages. Sea-level rise will shift shoreline morphology, reducing the areal extent of some habitats that are suitable for the industry. Changes in rainfall patterns will increase the turbidity and nutrient loading of rivers, potentially triggering harmful algal blooms and negatively affecting bivalve farming. In addition, ocean acidification may disrupt the early developmental stages of shellfish.

Some of the most damaging but least predictable effects of climate change relate to the emergence, translocation and virulence of diseases, parasites and pathogens, although parasites and diseases in finfish aquaculture may be controlled through intervention. The spread of nuisance and non-native species is also potentially damaging.

Rising temperatures may create the opportunity to rear warmer water species in theUKand Ireland. Market forces, rather than technical feasibility, are likely to determine whether existing farmed species are displaced by new ones.

abstract PREPRINT at Warwick Taylor, M., Simon, P.L., Green, D.M., House, T. & Kiss, I.Z. (2011). From Markovian to pairwise epidemic models and the performance of moment closure approximations. Journal of Mathematical Biology, 64: 1021-1042.
Many if not all models of disease transmission on networks can be linked to the exact state-based Markovian formulation. However the large number of equations for any system of realistic size limits their applicability to small populations. As a result, most modelling work relies on simulation and pairwise models. In this paper, for a simple SIS dynamics on an arbitrary network, we formalise the link between a well known pairwise model and the exact Markovian formulation. This involves the rigorous derivation of the exact ODE model at the level of pairs in terms of the expected number of pairs and triples. The exact system is then closed using two different closures, one well established and one that has been recently proposed. A new interpretation of both closures is presented, which explains several of their previously observed properties. The closed dynamical systems are solved numerically and the results are compared to output from individual-based stochastic simulations. This is done for a range of networks with the same average degree and clustering coeffcient but generated using different algorithms. It is shown that the ability of the pairwise system to accurately model an epidemic is fundamentally dependent on the underlying large-scale network structure. We show that the existing pairwise models are a good fit for certain types of network but have to be used with caution as higher-order network structures may compromise their effectiveness.
abstract PREPRINT as PDF Supplementary Information Green, D.M., Werkman, M. & Munro, L.A. (2012) The potential for targeted surveillance of live fish movements in Scotland. Journal of Fish Diseases 35: 29-37.
The network structure of the movements of live fish in the Scottish aquaculture industry has recently been demonstrated for 2003. In this paper, we enlarge this analysis to a longer three-year period from 2002 to 2004, the new data allowing complete coverage of at least one production cycle. The resulting network contains slightly more sites than that for a single year, and is denser with more arcs (directed site-to-site connections) present, but otherwise features recognisable in the one-year network are still recognisable in the three-year network. Arc removal algorithms (a proxy for targeted surveillance) were identified that could successfully reduce the portion of the network reachable from a node (a proxy for potential epidemic size) by approximately one third by removing as few as four arcs. This results from the high centrality of particular nodes and arcs. A strong community structure was identified in the network, corresponding with species farmed but only weakly geographical, with a high proportion of arcs occurring between management areas and catchments.


abstract Stirling STORRE Werkman, M., Green, D.M., Munro, L.A., Murray, A.G. & Turnbull, J.F. (2011). Seasonality and heterogeneity of live fish movements in Scottish fish farms. Diseases of Aquatic Organisms 96: 69-82
Movement of live animals is a key contributor to disease spread. Farmed Atlantic salmon Salmo salar, rainbow trout Onchorynchus mykiss and brown/sea trout Salmo trutta are initially raised in freshwater (FW) farms; all the salmon and some of the trout are subsequently moved to seawater (SW) farms. Frequently, fish are moved between farms during their FW stage and sometimes during their SW stage. Seasonality and differences in contact patterns across production phases have been shown to influence the course of an epidemic in livestock; however, these parameters have not been included in previous network models studying disease transmission in salmonids. In Scotland, farmers are required to register fish movements onto and off of their farms; these records were used in the present study to investigate seasonality and heterogeneity of movements for each production phase separately for farmed salmon, rainbow trout and brown/sea trout. Salmon FW-FW and FW-SW movements showed a higher degree of heterogeneity in number of contacts and different seasonal patterns compared with SW-SW movements. FW-FW movements peaked from May to July and FW-SW movements peaked from March to April and from October to November. Salmon SW-SW movements occurred more consistently over the year and showed fewer connections and number of repeated connections between farms. Therefore, the salmon SW-SW network might be treated as homogenous regarding the number of connections between farms and without seasonality. However, seasonality and production phase should be included in simulation models concerning FW-FW and FW-SW movements specifically.
abstract Stirling STORRE Green, D.M., Werkman, M., Munro, L.A., Kao, R.R., Kiss, I.Z., & Danon, L. (2011). Tools to study trends in community structure: application to fish and livestock trading networks. Prev. Vet. Med. 99: 225-228.
Partitioning of contact networks into communities allows groupings of epidemiologically related nodes to be derived, that could inform the design of disease surveillance and control strategies, e.g. contact tracing or design of `firebreaks' for disease spread. However, these are only of merit if they persist longer than the timescale of interventions. Here, we apply different methods to identify concordance between network partitions across time for two animal trading networks, those of salmon in Scotland (2002-4) and livestock in Great Britain (2003-4). Both trading networks are similar in that they moderately agree over time in terms of their community structures, but this concordance is higher -- and therefore community structure is more consistent -- when only the 'core' network of nodes involved in trading over the whole time series is considered. In neither case was higher agreement found between partitions close together in time. These measures differ in their absolute values unless appropriate standardisation is applied. Once standardised, the measures gave similar values for both network types.
abstract Stirling STORRE Soares, S., Green, D.M., Turnbull, J.F., Crumlish, M. & Murray, A. (2011). A baseline method for benchmarking mortality losses in Atlantic salmon (Salmo salar) production. Aquaculture, 314: 7-12.
On-farm databases provide a large diversity of information regarding fish health and stock performance. Mortality records held in on-farm database are indicators of fish health status and of great interest for studying fish health, such as patterns of diseases. Mortality records from a Scottish Atlantic salmon production database of one company were used to develop a method of benchmarking production losses due to mortality. The records used concerned mortality loss numbers of Atlantic salmon in the seawater phase. The median, 10th and 90th percentiles of mortality were calculated for each week of production from 88 production recorded cycles. These values were used to delimit the range of a standard mortality curve through the production cycle. The effects of the different mortality losses from each cycle on production in terms of costs and time consumed were also described. Likewise, substantial interannual variation in mortality time series is described as well as the mortality variation associated with three diseases (Pancreas Disease, Cardiomyopathy Syndrome and Infectious Pancreatic Necrosis).
abstract Stirling STORRE Werkman, M., Green, D.M., Murray, A.G. & Turnbull, J.F., (2011). The effectiveness of fallowing strategies in disease control in salmon aquaculture. Prev. Vet. Med. 98: 64-73.
Salmon production is an important industry in Scotland, with a retail value estimated at over £1 billion. However, this salmon industry can be threatened by the invasion and spread of diseases. To reduce this risk, the industry is divided into management areas that are physically separated from each other. Pathogens can spread between farms by local processes such as water movement or by long-distance processes such as live fish movements. Here, network modelling was used to investigate the significance of transmission routes at these two scales. We used different disease transmission rates (β), where infected farms had the probability of 0.10, 0.25 or 0.50 per month to infect each contacted farm. Interacting farms were modelled in such a way that neighbours within a management area could infect each other, resulting in two contacts per farm per month. In addition, non-local transmission occurred at random. Salmon are input to marine sites where they are raised to harvest size, the site is then fallowed; in the model the effects of different fallowing strategies (synchronised, partial synchronised and unsynchronised fallowing at the management area level) on the emergence of diseases were investigated. Synchronised fallowing was highly effective at eradicating epidemics when transmission rate is low (β = 0.10) even when long distance contacts were fairly common (up to 1.5 farm-1 month-1). However for higher transmission rates, long distance contacts have to be kept at far lower levels (0.15 contacts month-1 where ) when synchronised fallowing was applied. If fallowing was partially synchronised or unsynchronised then low rates of long-distance contact are required (0.75 or 0.15 farm-1 month-1) even if β = 0.10. These results demonstrate the potential benefits of having epidemiologically-isolated management areas and applying synchronised fallowing.
abstract Stirling STORRE Turnbull, J.F., Berrill, I., Green, D.M., Kaye, R., Morris, D., Murray, A., del-Pozo, J. & Shinn, A. (2011). Applied aquatic animal epidemiology in the UK. Aquaculture Research 42S1: 21-27.
This paper is a brief introduction to epidemiology and its application to farmed fish health and welfare with examples from the United Kingdom. Epidemiology has the potential to do a great deal more than just identify risk factors. Indeed in many cases useful risk factors cannot be identified due to the complexity of the disease problems and the lack of resources. Epidemiological principles or analytical techniques have been applied in animal welfare studies, and they can reduce the cost of disease monitoring or surveillance and disease control. However, for epidemiological studies to make a real contribution to farmed fish health and welfare it is often necessary to use multidisciplinary teams, obtain good data and coordinate efforts on the major problems.


abstract Stirling STORRE Sturm, A., Bron, J., Green, D.M., & Bury, N. (2010). Mapping of AF1 transactivation domains in duplicated rainbow trout glucocorticoid receptors. J. Mol. Endocrinology, 45:391-404.
The glucocorticoid receptor (GR) is a ligand-dependent transcription factor mediating the genomic effects of glucocorticoids. Two activation functions (AFs) are present in the GR. While the N-terminal AF1 is ligand independent, the C-terminal AF2 overlaps with the ligand-binding domain and is ligand dependent. In this study, we have mapped AF1 in duplicated rainbow trout GRs, called rtGR1 and rtGR2, showing a limited homology (24.5%) in the N-terminal domain. Ablation of this domain from rtGR1 or rtGR2 resulted in a marked decrease (>97%) in maximal hormone-dependent transactivation, but did not affect dexamethasone-binding activity or expression levels. This suggested that, similar to the situation in the human GR (hGR), AF1 is the main AF in the trout GRs. Sequence alignments with hGR suggested a localisation of AF1 to residues 70–230 of rtGR1 and 1–119 of rtGR2. These assignments were generally confirmed in the transactivation experiments with rtGR1- and rtGR2-derived mutants showing partial deletions of their N-terminal domains. In dexamethsone-treated cells (10-7 M, 2 h), the subcellular distribution of rtGR1 and rtGR2 mutants lacking the entire N-terminal domain, as well that of an rtGR1 mutant lacking the most N-terminal 234 amino acids, was similar to that of the corresponding wild-type GRs, suggesting that the disruption of transactivation activity was not caused by impairment of nuclear access of the mutants. Bioinformatic analyses predicted the presence of potential helical segments in the core of AF1 of rtGR1 and rtGR2, and further revealed that AF1 in rtGR1, rtGR2, and hGR shares a motif composed of hydrophobic and acidic amino acids.
abstract PREPRINT as PDF Green, D.M. and Kiss, I.Z. (2010). Large-scale properties of clustered networks: Implications for disease dynamics. Journal of Biological Dynamics, 4: 431-445.
We consider previously proposed procedures for generating clustered networks and investigate how these procedures lead to differences in network properties other than clustering. We interpret our findings in terms of the effect of the network structure on disease outbreak threshold and disease dynamics. To generate null-model networks for comparison, we implement an assortativity-conserving rewiring algorithm that alters the level of clustering while causing minimal impact on other properties. We show that many theoretical network models used to generate networks with a particular property often lead to significant changes in network properties other than that of interest. For high levels of clustering, different procedures lead to networks that differ in degree heterogeneity and assortativity, and in broader-scale measures such as R0 and the distribution of shortest path lengths. Hence, care must be taken when investigating the implications of network properties for disease transmission or other dynamic process that the network supports.
abstract Stirling STORRE Del-Pozo, J., Crumlish, M., Ferguson, H.W., Green, D.M., and Turnbull, J.F. (2010). A prospective longitudinal study of "Candidatus arthromitus"-associated rainbow trout gastroenteritis in the UK. Preventive Veterinary Medicine, 94: 289-300.
Rainbow trout gastroenteritis (RTGE) is an emerging disease of farmed rainbow trout (Oncorhynchus mykiss) reported in Croatia, France, Italy, Spain and the UK. The impact of RTGE at the site-level varies, and daily mortalities of 0.5–1% are common. Gross lesions include severe enteritis with congestion and oedema, and the segmented filamentous bacterium "Candidatus arthromitus" has been suggested as a possible aetiological agent. This paper presents findings from a prospective longitudinal epidemiology study that took place from June to November 2006 in 12 RTGE-positive UK sites, examining the impact and presentation of the disease, spread pattern, risk factors for disease presence and severity and the efficacy of current control strategies. RTGE was present in a total of 164 production units across the 12 sites, representing 39% of all stocked units. Fish losses due to RTGE totalled 61.4 tonnes. The plotting of epidemic curves revealed a propagating epidemic pattern in 51% affected units, suggesting RTGE is infectious. This was supported by risk analysis of fish transfers and site layouts, which revealed an increased risk of becoming RTGE positive after contiguity or fish transfer to and from a previously affected unit. Conditional multivariate logistic regression analysis identified eight variables significantly associated with the presence of RTGE. Seven variables were risk factors, including higher stocking densities, mortalities due to predation or handling, higher water temperature, higher daily feed input and contiguity or fish movement to and from a previous case. One variable was protective: the absence of an aeration system in a unit. General linear modelling identified a significant association of mean feed input per fish during an outbreak with RTGE cumulative mortality. Finally, great variability was observed with regards to the control strategies targeted to RTGE in these sites, suggesting an absence of a common strategy. In feed NaCl treatments apparently presented a palliative effect if given to affected fish during the clinical outbreaks, but they did not prevent recurrence.
abstract PREPRINT as PDF Green, D.M. (2010). A strategic model for epidemic control in aquaculture. Preventive Veterinary Medicine, 94: 119-127.
A stochastic metapopulation model of infectious disease was developed to model the spread of disease within and between sites of a region of an aquaculture industry. The study was a theoretical one examining the effect of transmission parameters through a sensitivity analysis. Production was modelled as either dispersed over many sites, or concentrated into small areas to provide "firebreaks" between such areas as a disease control strategy. The effectiveness of such a control strategy could then be examined for different industry and disease parameters (for example, overall production, and rates of within- and between-site infection). At the within-site level, contact was modelled as either frequency or density dependent, either of these extreme formulations being potentially appropriate for different diseases. Under density dependence, the effect of high host density of increasing the basic reproduction number R0 dominates, in contrast to the frequency-dependent model. However, for both model types, concentration of production into separate areas successfully slows the spread of simulated disease, particularly where long-distance transmission of the pathogen is weak due to fast attenuation of infectious agent over distance and time.
abstract PREPRINT as PDF Farkas, J.Z., Green, D.M. and Hinow, P. (2010). Semigroup analysis of structured parasite populations. Mathematical Modelling of Natural Phenomena, 5: 94-114.
Motivated by structured parasite populations in aquaculture we consider a class of size-structured population models, where individuals may be recruited into the population with distributed states at birth. The mathematical model which describes the evolution of such a population is a first-order nonlinear partial integro-differential equation of hyperbolic type. First, we use positive perturbation arguments and utilise results from the spectral theory of semigroups to establish conditions for the existence of a positive equilibrium solution of our model. Then, we formulate conditions that guarantee that the linearised system is governed by a positive quasicontraction semigroup on the biologically relevant state space. We also show that the governing linear semigroup is eventually compact, hence growth properties of the semigroup are determined by the spectrum of its generator. In the case of a separable fertility function, we deduce a characteristic equation, and investigate the stability of equilibrium solutions in the general case using positive perturbation arguments.


abstract PREPRINT as PDF Supplementary Information Green, D.M., Gregory, A. and Munro, L.A. (2009). Small- and large-scale network structure of live fish movements in Scotland. Preventive Veterinary Medicine, 91: 261-269.
Networks are increasingly being used as an epidemiological tool for studying the potential for disease transmission through animal movements in farming industries. We analysed the network of live fish movements for commercial salmonids in Scotland in 2003. This network was found to have a mixture of features both aiding and hindering disease transmission, hindered by being fragmented, with comparatively low mean number of connections (2.83), and low correlation between inward and outward connections (0.12), with moderate variance in these numbers (coefficients of dispersion of 0.99 and 3.12 for in and out respectively); but aided by low levels of clustering (0.060) and some non-random mixing (coefficient of assortativity of 0.16). Estimated inter-site basic reproduction number R0 did not exceed 2.4 at high transmission rate. The network was strongly organised into communities, resulting in a high modularity index (0.82). Arc (directed connection) removal indicated that effective surveillance of a small number of connections may facilitate a large reduction in the potential for disease spread within the industry. Useful criteria for identification of these important arcs included degree- and betweenness-based measures that could in future prove useful for prioritising surveillance.
abstract PREPRINT as PDF Supplementary Information Green, D.M. (2009). Coevolution of dispersal in a parasitoid-host system.Population Ecology, 51:253-260.
Interspecific interactions and the evolution of dispersal are both of interest when considering the potential impact of habitat fragmentation on community ecology, but the interaction between these processes is not well studied. We address this by considering the coevolution of dispersal strategies in a host–parasitoid system. An individual-based host–parasitoid metapopulation model was constructed for a patchy environment, allowing for evolution in dispersal rates of both species. Highly rarefied environments with few suitable patches selected against dispersal in both species, as did relatively static environments. Provided that parasitoids persist, all the variables studied led to stable equilibria in dispersal rates for both species. There was a tendency toward higher dispersal rates in parasitoids because of the asymmetric relationships of the two species to the patches: vacant patches are most valuable for hosts, but unsuitable for parasitoids, which require an established host population to reproduce. High host-dispersal rate was favoured by high host population growth rate, and in the parasitoid by high growth rates in both species.


abstract FULL TEXT as PDF Kiss, I.Z, and Green, D.M. (2008). Comment on "Properties of highly clustered networks".Physical Review E, 78: 048101.
We consider a procedure for generating clustered networks previously reported by Newman [Phys. Rev. E 68, 026121 (2003)]. In the same study, clustered networks generated according to the proposed model have been reported to have a lower epidemic threshold under susceptible-infective-recovered-type network epidemic dynamics. By rewiring networks generated by this model, such that the degree distribution is conserved, we show that the lower epidemic threshold can be closely reproduced by rewired networks with close to zero clustering. The reported lower epidemic threshold can be explained by different degree distributions observed in the networks corresponding to different levels of clustering. Clustering results in networks with high levels of heterogeneity in node degree, a higher proportion of nodes with zero connectivity, and links concentrated within highly interconnected components of small size. Hence, networks generated by this model differ in both clustering and degree distribution, and the lower epidemic threshold is not explained by clustering alone.
abstract FULL TEXT as PDF Green, D.M., Kiss, I.Z., Mitchell, A.P., and Kao, R.R. (2008). Estimates for Local and movement-based transmission of bovine tuberculosis in British Cattle. Proc R Soc B, 275: 1001-1005.
Both badgers and livestock movements have been implicated in contributing to the ongoing epidemic of bovine tuberculosis (BTB) in British cattle. However, the relative contributions of these and other causes are not well quantified. We used cattle movement data to construct an individual (premises)-based model of BTB spread within Great Britain, accounting for spread due to recorded cattle movements and other causes. Outbreak data for 2004 were best explained by a model attributing 16% of herd infections directly to cattle movements, and a further 9% unexplained, potentially including spread from unrecorded movements. The best-fit model assumed low levels of cattle-to-cattle transmission. The remaining 75% of infection was attributed to local effects within specific high-risk areas. Annual and biennial testing is mandatory for herds deemed at high risk of infection, as is pre-movement testing from such herds. The herds identified as high risk in 2004 by our model are in broad agreement with those officially designated as such at that time. However, border areas at the edges of high-risk regions are different, suggesting possible areas that should be targeted to prevent further geographical spread of disease. With these areas expanding rapidly over the last decade, their close surveillance is important to both identify infected herds quickly, and limit their further growth.
abstract FULL TEXT as PDF Kiss, I.Z., Green, D.M., and Kao, R.R. (2008). The effect of network mixing patterns on epidemic dynamics and the efficacy of disease contact tracing. Journal of the Royal Society Interface 5: 791-799.
In networks, nodes may preferentially contact other nodes with similar (assortatively mixed) or dissimilar (disassortatively mixed) numbers of contacts. Different patterns of contact support different epidemic dynamics, potentially affecting the efficacy of control measures such as contact tracing, which aims to identify and isolate nodes with infectious contacts. We used stochastic simulations to investigate the effects of mixing patterns on epidemic dynamics and contact-tracing efficacy. For uncontrolled epidemics, outbreaks occur at lower infection rates for more assortatively mixed networks, with faster initial epidemic growth rate and shorter epidemic duration than for disassortatively mixed networks. Contact tracing performs better for assortative mixing where epidemic size is large and tracing rate low, but it performs better for disassortative mixing at higher contact rates. For assortatively mixed networks, disease spreads first to highly connected nodes, but this is balanced by contact tracing quickly identifying these same nodes. The converse is true for disassortative mixing, where both disease and tracing are less likely to target highly connected nodes. For small epidemics, contact tracing is more effective on disassortative networks due to the greater resilience of assortative networks to link removal. Multi-step contact tracing is more effective than single-step tracing for assortative mixing, but this effect is smaller for disassortatively mixed networks.


abstract FULL TEXT as PDF Green, D.M., del Rio Vilas, V.J., Birch, C.P.D, Johnson, J., Kiss, I.Z., McCarthy, N.D. & Kao, R.R. (2007). Demographic risk factors for classical and atypical scrapie in Great Britain. Journal of General Virology, 88: 3486-3492.
Following the bovine spongiform encephalopathy (BSE) crisis, the European Union has introduced policies for eradicating transmissible spongiform encephalopathies (TSEs), including scrapie, from large ruminants. However, recent European Union surveillance has identified a novel prion disease, ‘atypical’ scrapie, substantially different from classical scrapie. It is unknown whether atypical scrapie is naturally transmissible or zoonotic, like BSE. Furthermore, cases have occurred in scrapie-resistant genotypes that are targets for selection in legislated selective breeding programmes. Here, the first epidemiological study of British cases of atypical scrapie is described, focusing on the demographics and trading patterns of farms and using databases of recorded livestock movements. Triplet comparisons found that farms with atypical scrapie stock more sheep than those of the general, non-affected population. They also move larger numbers of animals than control farms, but similar numbers to farms reporting classical scrapie. Whilst there is weak evidence of association through sheep trading of farms reporting classical scrapie, atypical scrapie shows no such evidence, being well-distributed across regions of Great Britain and through the sheep-trading network. Thus, although cases are few in number so far, our study suggests that, should natural transmission of atypical scrapie be occurring at all, it is doing so slowly.
abstract FULL TEXT as PDF Kao, R.R, Green, D.M., Johnson, J. & Kiss, I.Z. (2007). Disease dynamics over very different timescales: FMD and scrapie on the network of livestock movements in the UK. Journal of the Royal Society Interface, 4: 907-916.
We analyse the relationship between the network of livestock movements in the UK and the dynamics of two diseases: foot-and-mouth disease (FMD), which has an incubation period of days, and scrapie, which incubates over years. For FMD, the time-scale of expected epidemics is similar to the time-scale of the evolution of the network. We argue that, under appropriate conditions, a static network analysis can be an appropriate tool for gaining insights into disease dynamics even when the relevant time-scales are similar, as with FMD. We show that a subclass of 'linkage moves' maintains the network structure, and so removing these links has a dramatic effect on the number of potentially infected farms, an effect corroborated by simulations. In contrast, because scrapie has a low probability of transmission per contact and a long incubation period, a static network representation is probably appropriate; however, the signature of the network in the pattern of transmission is likely to be faint. Scrapie-notifying farms were more likely to be associated with each other via trading at markets than were control farms; however, network community structure proves to be less representative of prevalence patterns than geographical region. These contradictory indicators emphasize that appropriate observation time frames and good discrimination among types of potentially infectious contacts are vital in order for network analysis to be a valuable epidemiological tool.
abstract FULL TEXT as PDF Green, D.M. & Kao, R.R. (2007). Data quality of the Cattle Tracing System in Great Britain. The Veterinary Record, 161, 439-443
The Cattle Tracing System (CTS) of Great Britain was examined to investigate the quality of its data in terms of known errors and omissions, and their distribution. The proportions of erroneous or missing data have decreased steadily over time, with a marked improvement in the quality of the data since 2001, when recording became mandatory. There is little variation between regions in the quality of the data, but there are potentially important variations between the types of agricultural premises that cattle move to and from, and in other factors correlated with the type of premises, such as the age of the animals and the number of animal movements associated with the premises.
abstract FULL TEXT as PDF Parsons, D.J., Green, D.M., Schofield, C.P. and Whittemore, C.T. (2007). Real-time control of pig growth through an integrated management system. Biosystems Engineering 96: 257-266.
This paper describes the development and testing of the first prototype closed-loop, model-based, real-time system for the integrated control of pig growth and pollutant emissions. In each of two trials, growing pigs were reared from 30–50 to 65–125 kg in groups of 12 in 12 separate pens under controlled environment conditions at ADAS Terrington (Norfolk, England). They were fed ad libitum diets in which the protein content was controlled for each pen. Weight, estimated by visual image analysis, and feed intake were recorded daily for each pig. The control system was based on a mechanistic growth model. Each week, two model parameters were optimised using the data to improve the prediction, then the diet for each pen was optimised by adjusting the crude protein content between 140 and 190 g/kg [dry matter] to minimise the model error from a target for weight or fat depth. Part of the trial set weight gain targets of 50 and 60 kg over 70 days using two pens for each target. In three of the four pens the final mean weight of the pigs was within 2 kg of the target; in the fourth, growth was on target until it was interrupted close to the end of the trial. This trial has demonstrated the potential of the system to control the growth rate of pigs and has given encouraging but not conclusive results for the control of back fat depth.


abstract FULL TEXT as PDF Green, D.M., Kiss, I.Z., & Kao, R.R. (2006). Parasite strain coexistence in a heterogeneous host population. Oikos, 115: 495-503.
Heterogeneity in host susceptibility and transmissibility to parasite attack allows a lower transmission rate to sustain an epidemic than is required in homogeneous host populations. However, this heterogeneity can leave some hosts with little susceptibility to disease, and at high transmission rates, epidemic size can be smaller than for diseases where the host population is homogeneous. In a heterogeneous host population, we model natural selection in a parasite population where host heterogeneity is exploited by different strains to varying degrees. This partitioning of the host population allows coexistence of competing parasite strains, with the heterogeneity-exploiting strains infecting the more susceptible hosts, in the absence of physiological tradeoffs and spatial heterogeneity, and even for markedly different transmission rates. In our model, intermediate-strategy parasites were selected against: should coexistence occur, an equilibrium is reached where strains occupied only the extreme ends of trait space, under appropriate conditions selecting for lower R0.
abstract FULL TEXT as PDF Green, D.M., Kiss, I.Z., and Kao, R.R. (2006). Modelling the initial spread of foot-and-mouth disease through animal movements. Proceedings of the Royal Society Series B, 273: 2729-2735.
Livestock movements in Great Britain (GB) are well recorded and are a unique record of the network of connections among livestock-holding locations. These connections can be critical for disease spread, as in the 2001 epidemic of foot-and-mouth disease (FMD) in the UK. Here, the movement data are used to construct an individual-farm-based model of the initial spread of FMD in GB and determine the susceptibility of the GB livestock industry to future outbreaks under the current legislative requirements. Transmission through movements is modelled, with additional local spread unrelated to the known movements. Simulations show that movements can result in a large nationwide epidemic, but only if cattle are heavily involved, or the epidemic occurs in late summer or early autumn. Inclusion of random local spread can considerably increase epidemic size, but has only a small impact on the spatial extent of the disease. There is a geographical bias in the epidemic size reached, with larger epidemics originating in Scotland and the north of England than elsewhere.
abstract FULL TEXT as PDF Green, D.M., Kiss, I.Z. and Kao, R.R. (2006).Parameterization of individual-based models: Comparisons with deterministic mean-field models. Journal of Theoretical Biology 239, 289-297.
The relating of deterministic, mean-field models into network models, where epidemic spread occurs between interconnected susceptible and infectious individuals or populations, requires careful consideration. Here, we discuss models that consider differently the manner in which contact rate and infectiousness change over time, with different algorithms suitable for different underlying processes. Though these models give coincidental results to the mean-field in the case of large, highly connected networks, the results when sparsely connected networks are considered may differ. Different subsets of the parameters from the mean-field epidemic (R0, generation time, infectiousness, etc.) are preserved in each case. Despite these differences, simulated epidemics generated under some model architectures are insensitive to the average degree of contact amongst nodes, k. Model-based estimates of k may be model dependent, and must therefore be viewed with caution.
abstract FULL TEXT as PDF Kiss, I.Z., Green, D.M., & Kao, R.R. (2006). The network of sheep movements within Great Britain: network properties and their implications for infectious disease spread.Journal of the Royal Society Interface 3: 669-677. doi:10.1098/rsif.2006.0129
During the 2001 foot and mouth disease epidemic in the UK, initial dissemination of the disease to widespread geographical regions was attributed to livestock movement, especially of sheep. In response, recording schemes to provide accurate data describing the movement of large livestock in Great Britain (GB) were introduced. Using these data, we reconstruct directed contact networks within the sheep industry and identify key epidemiological properties of these networks. There is clear seasonality in sheep movements, with a peak of intense activity in August and September and an associated high risk of a large epidemic. The high correlation between the in and out degree of nodes favours disease transmission. However, the contact networks were largely dissasortative: highly connected nodes mostly connect to nodes with few contacts, effectively slowing the spread of disease. This is a result of bipartite-like network properties, with most links occurring between highly active markets and less active farms. When comparing sheep movement networks (SMNs) to randomly generated networks with the same number of nodes and node degrees, despite structural differences (such as disassortativity and higher frequency of even path lengths in the SMNs), the characteristic path lengths within the SMNs are close to values computed from the corresponding random networks, showing that SMNs have ‘small-world’-like properties. Using the network properties, we show that targeted biosecurity or surveillance at highly connected nodes would be highly effective in preventing a large and widespread epidemic.
abstract FULL TEXT as PDF Kiss, I.Z., Green, D.M. and Kao, R.R. (2006). Infectious disease contact tracing in random and scale-free networks. Journal of the Royal Society Interface, 3: 55 - 62.
Contact tracing aims to identify and isolate individuals that have been in contact with infectious individuals. The efficacy of contact tracing and the hierarchy of traced nodes— nodes with higher degree traced first—is investigated and compared on random and scale-free (SF) networks with the same number of nodes N and average connection K. For values of the transmission rate larger than a threshold, the final epidemic size on SF networks is smaller than that on corresponding random networks. While in random networks new infectious and traced nodes from all classes have similar average degrees, in SF networks the average degree of nodes that are in more advanced stages of the disease is higher at any given time. On SF networks tracing removes possible sources of infection with high average degree. However a higher tracing effort is required to control the epidemic than on corresponding random networks due to the high initial velocity of spread towards the highly connected nodes. An increased latency period fails to significantly improve contact tracing efficacy. Contact tracing has a limited effect if the removal rate of susceptible nodes is relatively high, due to the fast local depletion of susceptible nodes.
abstract FULL TEXT as PDF Kao, R.R., Danon, L., Green, D.M., & Kiss, I.Z. (2006). Demographic structure and pathogen dynamics on the network of livestock movements in Great Britain Proc. R. Soc. B, 273: 1999 - 2007. doi:10.1098/rspb.2006.3505.
Using a novel interpretation of dynamic networks, we analyse the network of livestock movements in Great Britain in order to determine the risk of a large epidemic of foot-and-mouth disease (FMD). This network is exceptionally well characterized, as there are legal requirements that the date, source, destination and number of animals be recorded and held on central databases. We identify a percolation threshold in the structure of the livestock network, indicating that, while there is little possibility of a national epidemic of FMD in winter when the catastrophic 2001 epidemic began, there remains a risk in late summer or early autumn. These predictions are corroborated by a non-parametric simulation in which the movements of livestock in 2003 and 2004 are replayed as they occurred. Despite the risk, we show that the network displays small-world properties which can be exploited to target surveillance and control and drastically reduce this risk.
abstract FULL TEXT as PDF Kiss, I.Z., Green, D.M., and Kao, R.R. (2006). The effect of network heterogeneity and multiple routes of transmission on final epidemic size. Math Biosci, 203: 124 - 136. doi:10.1016/j.mbs.2006.03.002
Heterogeneity in the number of potentially infectious contacts amongst members of a population increases the basic reproduction ratio (R0) and markedly alters disease dynamics compared to traditional mean-field models. Most models describing transmission on contact networks only account for one specific route of transmission. However, for many infectious diseases multiple routes of transmission exist. The model presented here captures transmission through a well defined network of contacts, complemented by mean-field type transmission amongst the nodes of the network that accounts for alternative routes of transmission. The impact of these combined transmission mechanisms on the final epidemic size is investigated analytically. The analytic predictions for the purely mean-field case and the transmission through the network-only case are confirmed by individual-based network simulations. There is a critical transmission potential above which an increased contribution of the mean-field type transmission increases the final epidemic size while an increased contribution of the transmission through the network decreases it. Below the critical transmission potential the opposite effect is observed.


abstract FULL TEXT as PDF Kiss, I.Z., Green, D.M. and Kao, R.R. (2005). Disease contact tracing in random and clustered networks. Proc. Roy. Soc. B, 272, 1407–1414 doi:10.1098/rspb.2005.3092
The efficacy of contact tracing, be it between individuals (e.g. sexually transmitted diseases or severe acute respiratory syndrome) or between groups of individuals (e.g. foot-and-mouth disease; FMD), is difficult to evaluate without precise knowledge of the underlying contact structure; i.e. who is connected to whom? Motivated by the 2001 FMD epidemic in the UK, we determine, using stochastic simulations and deterministic ‘moment closure’ models of disease transmission on networks of premises (nodes), network and disease properties that are important for contact tracing efficiency. For random networks with a high average number of connections per node, little clustering of connections and short latency periods, contact tracing is typically ineffective. In this case, isolation of infected nodes is the dominant factor in determining disease epidemic size and duration. If the latency period is longer and the average number of connections per node small, or if the network is spatially clustered, then the contact tracing performs better and an overall reduction in the proportion of nodes that are removed during an epidemic is observed.
abstract FULL TEXT as PDF Green, D. M. and Whittemore, C. T. (2005). Calibration and sensitivity analysis of a model of the growing pig for weight gain and composition. Agricultural Systems 84(3):279-295.
The model concerned was a mechanistic model for pig growth and composition which was constructed using novel algorithms for the accumulation of protein and lipid. The model was parameterised using data collected from a serial slaughter trial of three different types of pigs and the sensitivity to parameters of biological importance was explored. The types ('Landrace' type, 'Pietrain' type and 'Meishan' type) were chosen to represent 'lean', 'meaty' and 'fatty' types.
After optimisation, fitted parameters were found to lie close to the values that would be expected given the assumptions made during the construction of the model, with a predicted maximum protein retention rate of 0.20 kg d-1 and an efficiency of recapture of amino acids during turnover of 0.94. Mean absolute percentage errors at slaughter point for modelled live weight were ca. 5%, compared with 8% for protein mass and 13% for lipid mass, suggesting that live weight gain is easier to predict than the partitioning of biomass into retained protein and lipid.
abstract FULL TEXT as PDF Doeschl-Wilson, A.B., Green, D.M., Fisher, A.V., Carroll, S., Schofield, C.P & Whittemore, C.T. (2005). The relationship between body dimensions of living pigs and their carcass composition. Meat Science 70:229-240.
The performance of a visual image analysis (VIA) system was tested with regards to its potential to determine in vivo carcasscomposition and conformation, either alone, or in conjunction with other in vivo measures such as live weight and backfat depth.Pigs of both sexes of a commercial type were reared and slaughtered at weights ranging from 50 to 120 kg. Feeding was ad libitumon diets ranging from 0.14 to 0.19 kg kg^-1 crude protein content to produce animals of a range of body condition. Two analyseswere carried out: the first analysis addressed the relationship between dimensionless carcass and VIA indices; the second analysisassessed the relationship between carcass composition and VIA body shape using detrended carcass and VIA data, which were producedby removal of allometric growth trends.
A statistically significant relationship (P < 0.05) between in vivo VIA body size and shape and carcass muscle dimensions andcomposition was found for most body regions. Adjusted R2 statistics ranged between 0.13 and 0.54 for relative fat weights andbetween 0.14 and 0.51 for relative lean weights. The predictive power of the regression models, indicated by R2-like statistics forprediction, was approximately 70% of the adjusted R2 values. The descriptive and predictive powers of the corresponding modelsgenerally strengthened if VIA indices were combined with other in vivo measurements. The relationships between in vivo and carcassmeasures remained statistically significant (P < 0.05) after removal of the growth trends, although adjusted R2 statistics generallydecreased. The predictive power of models corresponding to the detrended measures was, however, weak.
The results show in vivo VIA measurements to be useful in the estimation of muscle size, carcass conformation and composition,all of which are of significant importance to the pig production, marketing and processing industries.


abstract FULL TEXT as PDF Doeschl, A.B., Green, D.M., Whittemore, C.T., Schofield, C.P., Fisher, A.V. and Knap, P.W. (2004). The relationship between the body shape of living pigs and their carcass morphology and composition. Animal Science 79:73-83.
The conformation, tissue composition, and chemical composition of three types of pigs, given food ad libitum and slaughtered over a nominal live weight range of 35 to 115 kg, was assessed in relation to data provided on the live animals by a visual image analysis (VIA) system. The pig types were named as '¾ Landrace', '½ Pietrain', and '¼ Meishan' types, representing 'attenuated', 'blocky', and 'flabby' types. Three analyses of the shape, conformation and composition data were performed. First, the relationship between conformation and age/size was assessed using linear regression of logarithmically transformed VIA and carcass data. In relation to age, '½ Pietrain' pigs were found by both VIA and carcass measurements to have the widest shoulders. Both analyses also found this type to have the widest ham, trunk, and shoulders in relation to body length across most of the body length range studied, although the greatest rate of increase in ham width in relation to body length was found in the '¼ Meishan' type pigs. Second, the relationship between composition and VIA shape was examined using linear regression of transformed and standardized data. Significant relationships were found between fat, lipid, muscle, and protein weight and VIA shape, although relationships were weaker for protein and muscle weight. For fat and lipid, the VIA shape measures from the trunk region proved the most informative, whereas the VIA ham measures proved the most informative for muscle and protein. Third, detrended measures of composition/conformation and shape were used to remove the effect of animal size from the data. Removal of the variation due to growth generally led to substantial decreases in the adjusted R2 statistics and in the R2-like statistics for prediction. Although in the models without detrending, relative fat and lipid weight had been found most strongly correlated with VIA shape, relative muscle was found most strongly correlated with shape in the detrended data. This was considered to result from the low between-animal variation in the data set combined with greater across-weight variation in fat and lipid weights than muscle and protein weights in the data without trend removal. Future trials with greater between-animal variation imposed would allow more precise determination of the relationship between conformation and shape.
abstract FULL TEXT as PDF White, R. P., Schofield, C. P., Green, D. M., Parsons, D. J. and Whittemore, C. T. (2004). The effectiveness of a Visual Image Analysis (VIA) system for monitoring the performance of growing/finishing pigs. Animal Science 78:409-418.
A visual image analysis (VIA) system provided continuous, automatic collection of size and shape data for a total of 116 pigsslaughtered serially from 25 to 115 kg live weight. Males and females of three types of pigs ('Meishan' type, 'Pietrain' type,and 'Landrace' type) were selected to provide variation in both composition and conformation (the three types being, respectively,'fat', 'blocky', and 'lean'). Results below are presented in this order. Regression analysis was used to relate VIA size toplatform weigher (FIRE) measurements of live weight. Residual maximum likelihood (REML) analysis showed that at the observedgrowth rate, a change in pig state could be detected by VIA after 8, 9, and 10 days respectively for the three types, and bythe platform weigher system after 12, 4, and 13 days (in both cases with a confidence of 95%). Artificial neural network andcanonical variates analysis were used to test the ability of VIA to distinguish between pig types and sexes. With crossvalidation, the canonical variates analysis correctly classified the three types in 72, 83, and 64% of observations, and theneural network in 81, 81, and 64% of observations. The VIA system is considered to be a valuable monitoring system which mayplay a rôle in the construction of integrated management systems (IMS).


abstract FULL TEXT as PDF Whittemore, C. T., Green, D. M., Wood, J. D., Fisher, A. V. and Schofield, C. P. (2003). Physical and chemical composition of the carcass of three different types of pigs grown from 25 to 115 kg live weight. Animal Science 77: 235-245.
A total of 74 pigs representing three commercially available crossbred types, Landrace (50%), Pietrain (50%) andMeishan (25%), were given food ad libitum over a 25- to 115-kg growth period and serially slaughtered forphysical and chemical analysis in five groups at 32, 42, 63, 82 and 114 kg live weight (W). Results are presented inthe order of pig type as above. Pig types grew at similar overall rates of live body gain, but the Meishan type atemore food and had greater back fat depth. The Pietrain type was least fat. Dissected fatty tissue grew substantiallyfaster than the carcass as a whole; allometric exponents being 1.64, 1.34 and 1.52 (P < 0.05) for the Landrace,Pietrain and Meishan types respectively. Dissected lean tissue gains were 0.419, 0.427 and 0.308 kg daily(P < 0.01), and dissected fatty tissue gains were 0.251, 0.158 and 0.218 kg daily (P < 0.05); the Meishan type beingslowest for lean gain and the Pietrain type slowest for fatty tissue gain. The Pietrain type had the largest cross-sectional area of the longissimus dorsi muscle, and the Meishan type the smallest. The pelvic limb of the Meishantype lost density (as measured by specific gravity) fastest, and that of the Pietrain slowest as the pigs grew. TheMeishan type had a lower proportion of its carcass lean and a higher proportion of its carcass fat in the pelvic limbthan did the other two types. For each kg of live-weight gain, 0.037, 0.041 and 0.032 kg (P < 0.05) of chemicalprotein was deposited in the pelvic limb of the three types respectively. Equivalent values for chemical lipid were 0.041, 0.035 and 0.041 (P < 0.05). The Meishan type retained protein at a relatively slower rate in the pelvic limbthan in the body as a whole. The Pietrain type had the greatest ultimate protein mass in the pelvic limb. Estimationof whole body protein content as a linear function of pig live weight gives coefficients of 0.154, 0.178 and 0.168 kg (P < 0.05) for the three types respectively. Equivalent values for whole body lipid content were 0.269, 0.214 and0.274 (P < 0.05). Best estimates of the daily rates of protein retention in the body of the whole live pig were 0.152,0.197 and 0.142 kg/day for the Landrace, Pietrain and Meishan types respectively.
abstract FULL TEXT as PDF Fisher, A.V., Green, D.M., Whittemore, C.T., Wood, J.D. & Schofield, C.P. (2003). Growth of carcass components and its relations with conformation in pigs of three types. Meat Science, 65: 639-650.
The growth of carcass tissues and developmental changes in tissue distribution were studied in three pig types which were selected to have morphologies that could be described as attenuated, blocky and flabby. These were achieved by incorporating, respectively, Landrace, Pietrain and Meishan genes to give commercial 'types' (designated L, P and M,) exhibiting some of the phenotypic qualities of these breeds. Twenty-five female pigs of each type with an average start weight of 27.2 kg were fed ad libitum and slaughtered over a (nominal) live weight range of 35-115 kg. Relations were quantified using the logarithmic transformation of data in the allometric model. Significant type differences in relative growth rates (the b coefficient or slope in the logarithmic plot) were not common. occurring in 14% of the relationships examined, whereas 61% of differences in the constant term (a or intercept in the logarithmic plot) were significant. Increase in carcass weight with age was not different between the types but in relation to slaughter live weight (dressing percentage) P had the highest value, M the lowest. Carcass dimensions showed that, relative to body length, P had the widest ham and shoulder over the whole size range whereas M had a deep (ventral-dorsal) shoulder and wide belly, attributes of shape that would be regarded as undesirable by the meat trade. Relative to carcass weight, L was only slightly longer (3 mm) in the body than M at the heavy end of the weight range but markedly longer (29 mm) than P. These differences in carcass conformation were also evident in the shape of pelvic limb muscles which, at a given length, were lightest and narrowest in M and, in some cases, heavier and wider in P than in L. Pelvic limb volume relative to limb length was greatest in P and least in M. Carcass composition (at a given prepared side weight) of P was characterized by low fat and high lean weights, and a high lean to bone ratio. Subcutaneous fat b was lowest in P, indicating that P was early maturing, but M carcasses had the most subcutaneous and intermuscular fat, also indicative of early maturing. There was, thus, no simple relation between maturity characteristics and carcass composition. M carcasses also had the greatest weight of skin. P had a light hindloin and heavy pelvic limb, M a heavy shoulder. hindloin and flank but a light pelvic limb; L had a light shoulder but a heavy foreloin. The distribution of individual tissues mirrored these differences in joint weights to a varying extent; in most cases the match (relatively heavy/light) was with a simile tissue. These data, thus, indicated some pig type differences in tissue distribution and of particular significance was the relatively light lean mass in the pelvic limb of M. The results showed important differences in carcass quality between commercially available pig types differing in conformation.
abstract FULL TEXT as PDF Green, D.M. & Whittemore, C.T. (2003). Architecture of a harmonised model of the growing pig for the determination of dietary net energy and protein requirements and of excretions into the environment (IMS Pig). Animal Science 77: 113-130.
The model incorporates, amongst its novel components, variable efficiency coefficients in the simulation of the responses of growing pigs to nutrient inputs, and thereby increases the accuracy and efficacy of control of feeding and nitrate excretion. The model determines (rather than is presented with) net energy and required amino acid level and balance. The estimation of protein turn-over as a function of rate of protein retention, protein mass and the maturity of the pig was found to be central to both the energy (ATP) and protein economy. Protein turn-over varied from around 0.14 to 0.08 of the protein mass depending upon the size of the pig. Efficiencies of energy yield from lipid, starch (and sugar), protein and (fibre-derived) volatile fatty acids were calculated to be 0.98, 0.86, 0.56 and 0.58 for ATP production and 0.90, 0.70, 0.50, and 0.44 for lipid retention, respectively. The maximum efficiency of use of ileal digestible amino acids was determined as around 0.85. The energy cost of protein synthesis was equivalent to 4.2 MJ metabolizable energy (ME) per kg, and the efficiency of use of ME for protein retention varied from 0.55 to 0.40 depending on the protein mass of the pig. The components of the model and the biochemical drivers are described in detail, and proof of principle of the main elements is presented. The model is different in its architecture to other published simulation models, and is considered to add to the present knowledge base in this discipline.
abstract FULL TEXT as PDF Green, D.M., Brotherstone, S., Schofield, C.P. & Whittemore, C.T. (2003). Food intake and live growth performance of pigs measured automatically and continuously from 25 to 115 kg live weight. Journal of the Science of Food and Agriculture 83: 1150-1155.
Continuous automatic measurement and electronic recording provided data for individual pig live weight and food intake on a daily basis for a total of 74 pigs representing three commercially available crossbred types of mixed genetic origin, which were kept under commercial farming conditions to determine practical expectations for food intake and performance characteristics over the whole of the growth range from 25 to 115 kg live weight. The pig types were named as Landrace type, Pietrain type and Meishan type, having been chosen to represent lean, meaty and fatty types. The pigs were fed ad libitum and serially slaughtered in groups at 32, 42, 63, 82 and 112 kg live weight (W). Results are presented in the order of pig type as above. Daily live weight gains at 80 kg live weight were 0.93, 0.94 and 0.91 kg day-1 and at 100 kg were 1.04, 0.81 and 0.75 kg day-1. Food intake curves differed with respect to both position and shape, with Pietrain type pigs showing a lower increase with increasing live weight. Significant differences were found in the levels of fatness, as measured by P2 backfat depth, of the three pig types across the weight range examined (P < 0.001), in order of fatness Pietrain type < Landrace type < Meishan type. Copyright © 2003 Society of Chemical Industry


abstract FULL TEXT as PDF Whittemore, C.T., Green, D.M. (2002). The description of the rate of protein and lipid growth in pigs in relation to live weight. Journal of Agricultural Science. 138: 415-423.
The paper seeks to compare the effectiveness of various functions as appropriate to express protein retention and lipid retention as a function of pig live weight. Linear descriptors were adequate for protein growth over the 20 to 120 kg live weight range, but not adequate when higher live weights were included. Linear descriptors were not adequate for lipid growth over any weight range. Quadratic and cubic polynomials, and the logistic function, were faulted on their failure to describe the known biology. Augmentation of the allometric function added nothing to the simple form. The Gompertz function was consistently effective for the description of daily protein retention rate. The Bridges and Richards functions are more flexible than the Gompertz with respect to the point of inflection, but when applied they resembled the Gompertz and therefore did not materially influence the descriptive outcome. The cubic polynomial, augmented allometric, Bridges and Richards functions, although favoured in other reports, were found to add nothing to the more simple functions. It is concluded that protein retention can be well expressed in relation to live weight by linear and Gompertz functions. Lipid retention could be well described by an allometric relationship with pig mass.
abstract FULL TEXT as PDF Schofield, C.P., Whittemore, C.T., Green, D.M. & Pascual, M.D. (2002). The determination of beginning and end of period live weights in growing pigs. Journal of the Science of Food and Agriculture. 82: 1672-1675.
Optimisation of meat production systems depends upon accurate performance assessment. Weigh platforms in electronic feeding stations allow continuous monitoring of individual pig weight change. Such a system was compared with a conventional mechanical scale with the general result: feeding station weight (kg) = 1.02 (0.010) scale weight - 1.65 (0.766). As compared with the conventional scale measurement at start and at finish of test, back extrapolation of daily weights collected over time in electronic feeding stations may give an improved estimate of start weight, and forward extrapolation an improved estimate of final weight. Regression predictions suggest this to be the case for final weight, but for start weight both systems are open to error due to variation in weight in the early days of test. However, such variation (and its diagnostic utility) is only observable with a system that continuously monitors weight.


abstract FULL TEXT as PDF Whittemore, C.T., Green, D.M. & Knap, P.W. (2001). Technical review of the energy and protein requirements of growing pigs: food intake. Animal Science 73: 3-17.
Food intake in pigs is highly variable across different production circumstances. This report concludes from a critical review of published observations that it was unrealistic to expect from the scientific literature purporting to express nutrient requirement any reasonable prediction of the particular food intake of groups of pigs. None the less, such knowledge is essential for the practical purposes of their day- to-day nutrition. The literature does however yield general principles from which may be derived: (a) the likely forms (but not the parameter values) of intake functions relating food intake to pig live weight; and (b) the likely factors involved in the modulation of food intake at any given live weight. Using these principles two methods for determining on farm food intake from the use of simple and available records were proposed. The first requires knowledge only of start and final weight, the time elapsed, and total food intake: it involves two steps, the determination of a suitable growth curve followed by the fitting of a suitable food intake curve. The second method is appropriate in the absence of information on total food intake, and requires a minimum number of spot measurements through the growth period. Different functions were tested for the curve of best fit. As a further benefit it appeared that models could be constructed from the information presented that would speculate for diagnostic purposes upon the likely modulators of food intake. Such models could explore the constraints of gut capacity, the energetic requirements of maintenance and potential growth, the influence of excessive or inadequate environmental temperature, the quality of housing and stocking density.
abstract FULL TEXT as PDF Whittemore, C.T., Green, D.M. & Knap, P.W. (2001). Technical review of the energy and protein requirements of growing pigs: energy. Animal Science 73: 199-215.
A review of work reported in the literature was used to present quantitative descriptions of energy dispositioning in the growing pig. These are detailed in the text, which points to preferred values, as well as to anomalies and lacunae. The review was prepared with the objective of allowing from its content the inclusive and quantitative modelling of energy requirement. Requirement is approached as the sum of the component factors; maintenance, protein retention and lipid retention. Conventional expressions of maintenance requirement, as some function of pig mass, were found unconvincing in their variety of expression of coefficients and exponents. The review concluded that maintenance is properly related to protein turn- over, and thereby requires at least to include elements of concomitant protein metabolic activity. It was also judged that maintenance costs might be farm-specific. The energy requirements for activity, gaseous losses and disease were identified as important, but unsatisfactory in their quantification. Exploration of the energy costs of uncomfortable ambient temperatures suggested that whilst the responses of the pig are open to sophisticated and relatively exact calculation, the description of comfort remained inexact. The efficiency of retention of lipid by direct incorporation was high and may comprise a substantial proportion of the dietary lipid supply. There was little evidence of variation in the efficiency of utilization of metabolizable energy from carbohydrate for lipid retention. The linear-plateau paradigm for protein retention was adopted. The efficiency of utilization of energy for protein retention measured by a variety of approaches was found to be highly variable, prone to error and the literature confused. It was concluded that the efficiency of use of metabolizable energy for protein retention would be a function of at least: (a) the absorbed substrate being metabolized for the synthesis of body protein, (b) the rate of total protein tissue turn-over associated with the retention of newly accreted protein and not already accounted in the estimate of maintenance, (c) the mass of protein tissue involved in turn-over, and (d) the degree of maturity attained, and any influence maturity may have upon the rate of turn-over of total body protein. Algorithms for energy requirement are presented based upon protein turn-over and these appear to have some consistency with empirical findings.
abstract FULL TEXT as PDF Whittemore, C.T., Green, D.M. & Knap, P.W. (2001). Technical review of the energy and protein requirements of growing pigs: protein. Animal Science 73: 363-374.
A review of work reported in the literature was used to present quantitative descriptions of protein use in the growing pig. These are detailed in the text, which also points to preferred values, and to anomalies and lacunae. The review was prepared with the objective of allowing from its content the inclusive and quantitative modelling of amino acid requirement. Requirement was approached as the sum of the component factors: maintenance and protein retention. Ileal trite digestible protein and amino acid requirements are presented in a form consistent with that forwarded for energy. Thus both energy and protein elements can be conceptualized within a single coherent framework. Priority uses for absorbed amino acids were assumed to be (a) to support endogenous protein losses resultant from the passage of food and incomplete re-absorption prior to the terminal ileum, (b) to replace lost hair and skin, and (c) to cover the basic maintenance losses which will occur as a result of minimal protein turn-over even when protein retention is zero. The bulk of the protein requirement was directly linked to the daily rate of protein retention, for which the linear- plateau response was accepted. For determination of the maximum rate of protein retention the Gompertz function was proposed, although the use of a single value throughout the growth period was not dismissed. The balance of amino acids for protein retention is specified as different from that for maintenance. Central to the approach was the proposal that the inefficiency of use of Heal digested ideal protein, even when not supplied in excess, was an expression of protein losses occurring as a result of protein turn-over. The requirement for the satisfaction of the losses from protein turn-over occurring as a consequence of protein retention, and therefore additional to the requirements for maintenance, was identified. Quantfication was attempted with sufficient success to warrant its inclusion into requirement estimation. It was concluded that this element addressed previously inadequately explained protein utilization inefficiencies. Algorithms are presented based upon protein turn-over which appear to be consistent with enipirical findings.


abstract FULL TEXT as PDF Green, D.M., Kraaijeveld, A.R. & Godfray, H.C.J. (2000). Evolutionary interactions between Drosophila melanogaster and its parasitoid Asobara tabida. Heredity 85: 450-458.
Drosophila melanogaster is attacked by parasitoids that develop internally in the larva. They can defend themselves by a cellular immune response (host resistance), although this can be disabled by parasitoid countermeasures (parasitoid virulence). D. melanogaster and its parasitoids are an excellent system in which to study coevolution experimentally. We designed an experiment to compare changes in resistance and virulence in replicate populations of flies and parasitoids maintained together for approximately 10 fly (five parasitoid) generations. The experiment had three treatments each with three replicates: (A) no parasitoids (B) outbred parasitoids (C) partially inbred parasitoids. Host resistance increased in treatments B and C but there was no difference between these treatments. Parasitoid virulence appeared not to change during the experiment. Host larvae in treatments B and C fed at lower rates than those in A, evidence of a trade-off between resistance and larval competitive ability. We found no evidence for local adaptation, as hosts from the different replicates of treatment C performed no differently against parasitoids from the same and other replicates. Also, we found no evidence for the evolution of behavioural traits in the host that could lead to lower probabilities of being attacked. Comparing the evolution of host resistance in these seminatural settings with that in artificial selection experiments provides insight into how the conflicting selection pressures on host resistance interact.


abstract Green, D.M. (1997). A New Record and a New Species of Dohrniphora (Diptera: Phoridae) from Malaysia. Malayan Nature Journal 50: 159-165.
Dohrniphora malaysiae sp. n. (Phoridae) is described from Malaysia, and a new variety of Dohrniphora spriggsi Disney is reported.

Refereed conference abstracts

FULL TEXT as PDF Wright, A.J., Green, D.M., & Kao, R.R. (2007). Analysing the Risk of a Bovine Brucellosis Epidemic in Great Britain Using the Cattle Tracing Scheme: Have We Just Been Lucky So Far? Proceedings of the British Society of Animal Science 2007 p. 195.
FULL TEXT as PDF Doeschl-Wilson, A.B., Green, D.M., Fisher, A.V., Carroll, S., Schofield, C.P. & Whittemore, C.T. (2005). The relationship between body dimensions of living pigs and their carcass composition. Proceedings of the British Society of Animal Science 2005 p. 80.
FULL TEXT as PDF (draft) Green, D.M. & Whittemore, C.T. (2004). Calibration and sensitivity analysis of aharmonised model of the growing pig. Journal of Agricultural Science 142: 243-250 (all abstracts).
FULL TEXT as PDF Green, D.M., Parsons, D.J., Schofield, C.P. & Whittemore, C.T. (2004). Real-time control of pig growth through an integrated management system (IMS). Proceedings of the British Society of Animal Science 2004 p. 21.
Parsons, D.J., Schofield, C.P., Green, D.M., Whittemore, C.T., Carroll, S., Kay, R. (2004). Real-time control of pig growth through an integrated management system (IMS). Institute of Agricultural Engineers 2004, Leuven, Belgium, 09/04.
FULL TEXT as PDF White, R.P., Parsons, D.J., Schofield, C.P., Green, D.M. & Whittemore, C.T. (2003). Use of visual image analysis for the management of pig growth in size and shape. Proceedings of the British Society of Animal Science 2003 p. 101.
FULL TEXT as PDF Döschl, A.B., Whittemore, C.T., Green, D.M., Fisher, A.V. & Schofield, C.P. (2003). Use of visual image analysis for the description of pig growth in size and shape. Proceedings of the British Society of Animal Science 2003 p. 21.
FULL TEXT as PDF Green, D.M. & Whittemore, C.T. (2002). Description and validation of a harmonised model of the growing pig for the optimisation of the utilisation and excretion of nutrients. Proceedings of the British Society of Animal Science 2002 p. 26.
FULL TEXT as PDF Whittemore, C.T., Green, D.M., & Schofield, C.P. (2001). Nutrition management of growing pigs. In: Integrated management systems for livestock (Eds. Wathes, C.M., Frost, A.R., Gordon, F. & Wood, J.D.) BSAS Occasional Publication No. 28. BSAS, Edinburgh. 89-95.

Other conference and workshop abstracts

Green, D.M. & Munro, L.A. (2009). Network modelling for epidemic control and targeted surveillance. 14th EAFP International Conference. p25.
Parsons, D.J., Schofield, C.P., Green, D.M. & Whittemore, C.T. (2004). Real-time, model-based control of pig growth. European Workshop for Decision Problems in Agriculture and Natural Resources. Silsoe Research Institute 27-28 September 2004.

Books and book chapters

Green, D.M. & Parsons, D.J. (2006). The place of models in the new technologies of production systems. In: Mechanistic Modelling in Pig and Poultry Production (Eds. Fisher, C., Gous, R. & Morris, T). CABI Publishing, Wallingford, UK.
Green, D. Wellock, I. & Whittemore, C. (2006). Simulation modelling. In: Whittemore's Science and Practice of Pig Production (3rd edition; Eds Kyriazakis, I. & Whittemore, C.T.). Blackwell Publishing, Oxford, UK.
Whittemore, C.T. & Green, D.M. (2001). Growth of the young weaned pig. In: The Weaner Pig (Eds. Varley, M. & Wiseman, J.). CABI Publishing, Wallingford, UK.
abstract Green, D. M. (2000). Coevolutionary dynamics in a parasitoid-host system. PhD thesis, University of London.
There are many documented examples of rapid evolutionary change, indicating that ecological and evolutionary processes can operate simultaneously. Often, rapid evolutionary changes are a consequence of new inter-specific interactions resulting from human influence.
The process of reciprocal evolutionary change between species is referred to as coevolution. Parasitoids and their hosts provide model systems in which coevolution can be studied. Here, the interaction between the two species is particularly close. Since parasitoids kill their hosts, the effects of the interaction upon host and parasitoid fitness are easily determined. Hosts may have resistance mechanisms against the parasitoids, and parasitoids may have countermeasures against these resistance mechanisms (virulence). Evolution of both sets of traits is possible.
Described in this thesis are a number of experiments that investigate the interaction between the fruit fly, Drosophila melanogaster, and its parasitoids. Theoretical studies further analyse host - parasitoid coevolution, and investigate the effects of coevolution upon community structure.
The experimental work begins with an investigation into the coevolution between Drosophila melanogaster and Asobara tabida using laboratory population cages. An increase in host resistance was found on exposure to parasitism, which was correlated with lower larval feeding rate. No evidence was found for increased concealment behaviour by the host. No response to selection was found in the parasitoid.
Further experiments show a possible correlation between D. melanogaster resistance against Leptopilina boulardi and larval foraging behaviour. Another possible correlation exists between resistance against A. tabida and susceptibility to Pachycrepoideus vindemiae. Inbreeding in D. melanogaster was found to lead to a decrease in resistance against A. tabida.
Modelling demonstrated the importance of host resistance specificity in determining community structure. If resistance mechanisms act against all parasitoid species, then the persistence of multiple parasitoid species within the community is less likely than if resistance mechanisms against each parasitoid species are separate.

Trade press articles

Green, D.M. (2012). Lessons from ants and traffic jams: Complexity in aquaculture biosecurity. Fish Farming Xpert, May 2012 3: 43-46.
Murray, A.G., Soares, S., & Green, D.M. (2011). Benchmarking losses in salmon farms. Fish Farmer, May/June 2011, p. 44.
Green, D.M. (2011). Fishes, cages, sites, and countries: epidemic modelling in aquaculture. Fish Farmer, July/Aug 2011, 24-25.

Posters (a selection)

PDF Green, D.M. (2009). Epidemic Modelling for a Changing Industry. Poster at EAFP, Prague 2009.
PDF Green, D.M. & Kao, R.R. (2008). Analysis of M. bovis genotypes in GB and the implications for BTB control. Poster at SVEPM, Liverpool 2008.
PDF Green, D.M., del Rio Vilas, V., Birch, C. & Kao, R.R. (2007). Patterns of atypical scrapie in Great Britain. Poster at SVEPM, Espoo 2007.
PDF Green, D.M., Mitchell, A. & Kao, R.R. (2007). The impact of different transmission routes of bovine tuberculosis in British cattle. Poster at SVEPM, Espoo 2007.
PDF Green, D.M., Kiss, I.Z., & Kao, R.R. (2005). Analysis of FMD control strategies using contact network simulations. Poster at SVEPM, Nairn 2005.

Invited seminars

University of Strathclyde, Department of Mathematics and Statistics. November 2012.
University of Sussex, Department of Mathematics. May 2010.
University of Glasgow, Faculty of Veterinary Medicine. July 2009.
University of Saskatchewan, Department of Veterinary Medicine. May 2009.
Royal Statistical Society, Aberdeen local group. March 2009.
University of York, Department of Biology. December 2008.
University of Stirling, Department of Mathematics and Computer Science. February 2008.
Fisheries Research Services. January 2008.
Health Protection Agency. April 2007.
The 3rd International Symposium on Transmission Models for Infectious Diseases. Kyoto, Japan. January 2007.
Recent Advances in Pig and Poultry Modelling, Ithala Game Reserve, KwaZulu-Natal, South Africa, April 2005.
ZODIAC, University of Wageningen. September 2002.
Nutreco, Boxmeer, The Netherlands. September 2002.