群体饲喂条件下猪个体采食量的评估模型

在大多数生长试验中,试验动物往往成群地饲养在同一圈中。在试验过程中,有时有个别动物在生长速度上会表现出与其同伴较大的差异,或者死亡。出现此种情况时,如果不能为统计日平均采食量(average daily feed intake,ADFI)和饲料效率而对试验动物的圈采食量(penfeedintake,PFI)进行重新评估,那么试验数据集将丢弃该圈的观察值。因此,我们开发了一个模型,用来评估群体饲喂条件下猪的个体采食量(Individual feed intake,IFI),并通过群体饲喂模拟试验对该评估模型进行验证。在推荐模型中,将每个受影响圈的采食量(FeedIntake,FI)分为用于维持生存需要的采食量(FIm)和用于生长需要的采食量(FIg)两个部分。首先,试验期间个体猪的FIm按NRC(1998)所列出的维持生存所需代谢能的评估方法进行测算,然后将同圈内所有猪的FIm相加,即得到圈FIm。圈采食量(PFI)与圈FIm之差就是该圈动物用于生长所需的那部分采食量。接下来,把该差额部分的采食量平均分给圈内动物每单位增重上,即为FIg。最后,猪圈中被淘汰猪的IFI估测值即为所淘汰猪的FIm与FIg之和,原来的圈采食量(PFI)减去该FI估测值后即得剩余猪的新PFI。个体采食量评估的准确性取决于维持生存代谢能预测方程和饲料原料能量分析值的准确性。在模拟试验中,我们将推荐模型的精确度与2个其它模型的精确度作了比较。模拟试验1显示,在所有测试阶段中推荐模型的精确度至少高于其它2个模型中的任一种方法(P<0.001);模拟试验2也证明,推荐模型的精确度高于其他2种方法。由于IFI的人工计算相对麻烦,为此我们开发了饲料采食量校正电子数据表(feed intake correction spreadsheet,FICS),即一个利用宏指令可对FI进行校正的Excel电子数据表。FISC包括了推荐模型中的所有测算程序,用户可免费从指定的网页上下载Excel文件和使用说明书。     

Effect of selection for residual feed intake on feeding behavior and daily feed intake patterns in Yorkshire swine

Residual feed intake (RFI) is a measure of feed efficiency defined as the difference between observed and predicted feed intake based on average requirements for growth and maintenance. The objective of this study was to evaluate the effect of selection for decreased RFI on feeding behavior traits and to estimate their relationships with RFI. Three data sets from the 4th and 5th generations of a selection experiment with a line selected for reduced RFI (LRFI) and a randomly selected control line (CTRL) were analyzed. Lines were mixed in pens of 16 and evaluated for feeding behavior traits obtained from a single-space electronic feeder over a growing period of ~3 mo before ~115 kg. The following traits were evaluated as averages over the entire test period and over the first and second half of the test period: number of visits per day and hour; occupation time per day, visit, and hour; feed intake (FI) per day, visit, and hour; and FI rate per visit. Models used included fixed effects of line and feeder, covariates of on-test age and FI per day, and random effects of pen, on-test group, sire, and litter. Repeated measures models were used to analyze feeding patterns during the day. The LRFI pigs had significantly less FI per day than CTRL pigs for all 3 data sets. With adjustment for FI per day, line differences of all traits were in the same direction for all 3 data sets but differed in significance and size. Feed intake per visit and hour and visits per day and hour did not differ between lines, but the trend was for LRFI pigs to have fewer visits, in particular during peak eating times. The LRFI pigs had a greater feeding rate and less occupation time per day, visit, and hour than CTRL pigs, but this was not significant for all data sets. Correlations of RFI with FI per day and visit and visits per day were positive. Average daily gain was positively correlated with FI per day and visit and occupation time per visit but negatively correlated with visits per day. Feed intake per day was positively correlated with backfat. In conclusion, feed efficiency may be affected by FI behavior because selection for decreased RFI has resulted in pigs that spend less time eating and eat faster.     

Use of n-alkanes to estimate feed intake in ruminants: a meta-analysis

Precise techniques to estimate feed intake by ruminants are critical to enhance feed efficiency and to reduce greenhouse gas emissions and nutrient losses to the environment. Using a meta-analysis, we evaluated the accuracy of the n-alkane technique to predict feed intake in cattle and sheep and assessed the relationships between feed intake and fecal recovery (FR) of n-alkanes. The database was composed of 28 studies, including 129 treatments (87 and 42 for cattle and sheep, respectively) and 402 animals (232 cattle and 170 sheep) fed at troughs, from published studies. Relationships between observed (in vivo measurement) and predicted feed intake by C31:C32 and C32:C33 n-alkane pairs were evaluated by regression. Meta-regression addressed the relationships between the difference in FR of n-alkane pairs and the error in intake estimation, as well as the amount and duration of C32 n-alkane dosing. Regression of observed intake on n-alkane-based estimates revealed good relationships in cattle (adjusted R² = 0.99 for C31:C32, and adjusted R² = 0.98 for C32:C33; P < 0.0001) and in sheep (adjusted R² = 0.94 for C31:C32, and adjusted R² = 0.96 for C32:C33; P < 0.0001). FR of natural n-alkanes showed a coefficient of variation of about 15% and 16% for C31 and C33, respectively, in cattle. In sheep, the coefficient of variation was 8% and 14% for C31 and C33, respectively. The relationships between the difference of FR of n-alkane pairs and the error in feed intake estimation in cattle were characterized by an adjusted R² = 0.83 for C31:C32 (P < 0.0001) and adjusted R² = 0.93 for C32:C33 (P < 0.0001). In sheep, they were characterized by an adjusted R² = 0.69 for C31:C32 (P < 0.001) and adjusted R² = 0.76 for C32:C33 (P < 0.001). The n-alkane technique provided the reliability for estimating feed intake in cattle and sheep in barn experiments. The present meta-analysis demonstrated that without correction for differences in FR of n-alkane pairs, deviation in feed intake prediction would occur. However, further research is necessary to determine the relationship between the n-alkane dosing procedure (daily amount and duration of dosing) and FR of n-alkane.     

Technical note: Development of an enzyme-linked immunosorbent assay for the determination of florfenicol and thiamphenicol in swine feed

One polyclonal antibody against florfenicol and thiamphenicol was produced and a competitive ELISA was developed for the detection of florfenicol and thiamphenicol in swine feed. The ELISA gave a 50% inhibiting concentration of 1.02 ng/mL for florfenicol. For swine feed fortified with 0.05 to 3.0 mg/kg, the interassay recoveries of florfenicol and thiamphenicol ranged from 86.4 to 118.6%, whereas intraassay recoveries of both drug ranged from 90.1 to 126.5% with less than 15% CV. Results obtained from HPLC-tandem mass spectrometry indicated this ELISA procedure could be used as a convenient method for rapid screening of florfenicol and thiamphenicol in swine feed.     

Use of group records of feed intake to select for feed efficiency in rabbit

Models for genetic evaluation of feed efficiency ( FE ) for animals housed in groups when they are either fed ad libitum ( F ) or on restricted ( R ) feeding were implemented. Definitions of FE on F included group records of feed intake () and individual records of growth rate ( G_F ) and metabolic weight ( M_F ). Growth rate ( G_R ) as FE measurement on R was used. Data corresponded to 5,336 kits from a rabbit sire line, from 1,255 litters in 14 batches and 667 cages. A five‐trait mixed model (also with metabolic weight on R, M_R ) was implemented including, for each trait, the systematic effects of batch, body weight at weaning, parity order and litter size; and the random effects of litter, additive genetic and individual. A Bayesian analysis was performed. Conditional traits such as and were obtained from elements of additive genetics ( and ) or phenotypic ( and ) (co)variance matrices. In the first case, heritabilities were low (0.07 and 0.06 for and , respectively) but null genetic correlation between the conditional and conditioning traits is guaranteed. In the second case, heritabilities were higher (0.22 and 0.16 for and , respectively) but the genetic correlation between and was moderate (0.58). Heritability of G_R was low (0.08). This trait was negatively correlated with and of animals on F, which indicate a different genetic background. The correlation between G_R and G_F was also low to moderate (0.48) and the additive variance of G_F was almost four times that of G_R, suggesting the presence of a substantial genotype by feeding regimen interaction.     

A model to estimate the financial consequences of classical swine fever outbreaks: principles and outcomes

A model is presented aimed at a financial analysis of a Classical Swine Fever outbreak. Financial consequences are calculated for affected parties, including governments (EU and national), farms, and related industries in the production chain. The model can be used to calculate the losses of a real outbreak as well as of a simulated one. In this article, the model is applied to the 1997/1998 outbreak of Classical Swine Fever in the Netherlands. Results show that total financial consequences of the outbreak are US $2.3 billion. Consequential losses for farmers and related industries are US $423 million and US $596 million respectively. Budgetary consequences for governments include less than 50% of the total losses calculated by the model. The model can be adapted easily to suit other diseases and countries.     

Bayesian analysis of the effect of selection for residual feed intake on growth and feed intake curves in Yorkshire swine

Gompertz growth functions were fitted to longitudinal measurements of daily feed intake (DFI) and BW of 586 boars and 495 gilts from a selection experiment in Yorkshire pigs for residual feed intake (RFI). The selection experiment consists of a line selected for low residual feed intake (LRFI) for 5 generations and a randomly selected control line (CTRL). The objectives of this study were to use Bayesian methods to estimate genetic parameters of the Gompertz curve parameters for DFI and BW, to evaluate the effect of selection for reduced RFI on the Gompertz parameters and shape of curves for DFI and BW, and to develop methodology for quantifying genetic variation at the level of the original phenotypes for DFI and BW based on the Bayesian analysis of the nonlinear model. Separate analyses were done for boars and gilts and for BW and DFI. A hierarchical model was specified in 2 levels: in the first level, the Gompertz function was modeled for each pig, and at the second level, a 3-trait linear mixed model was fitted to the 3 Gompertz parameters (asymptotic value, inflection point, and decay parameter), with fixed effects of line by generation and random effects of additive genetic and environmental effects. Bayesian methods were used to combine the 2 levels of modeling. A total of 30,000 random samples of the posterior distributions after convergence of Markov chains were used for inference. Posterior means of heritability within the first level of the model for the asymptotic value, inflection point, and decay parameter for DFI were 0.74, 0.66, and 0.82 for boars and 0.79, 0.70, and 0.57 for gilts; corresponding estimates for BW were 0.64, 0.58, and 0.60 for boars and 0.46, 0.35, and 0.33 for gilts. For DFI, LRFI boars had a reduced mature DFI (2.91 vs. 3.20 kg/d) and an earlier inflection point (85 vs. 95 d) compared with CTRL boars. For BW, LRFI boars had a lighter mature BW (279 vs. 317 kg), an earlier inflection point (184 vs. 198 d), and a decreased decay parameter (127 vs. 134 d) compared with CTRL boars. In contrast, LRFI gilts had a later inflection point (225 vs. 200 d) and a greater decay parameter (172 vs. 143 d) than CTRL gilts for BW. The other Gompertz curve parameters for DFI and BW for boars and gilts were considered not different between lines, with posterior probabilities of the line differences being greater than zero ranging from 0.1 to 0.9.     

A model for predicting feed intake of growing animals during exposure to pathogens

A general model is proposed for predicting the effects of subclinical pathogen challenges of different doses and virulence on the relative feed intake (RFI) of animals. The RFI is defined as the feed intake (FI, kg/d) of the animal challenged by a pathogen divided by its FI in the same state had it not been challenged. Actual FI can be predicted from the RFI and the animal's state. The RFI was assumed to be affected only when animals were na?ve to a particular pathogen (i.e., had not previously experienced it) and when the challenge dose was above a predetermined threshold. The model is for the period from recognition of a pathogen through acquisition and subsequent expression of immunity. The way in which RFI changes with time is described by 5 main parameters and is based on data for RFI during different pathogen challenges of a range of hosts. Lag time (L, d) is the delay from a pathogen challenge until any effects on RFI are seen. Reduction time (R, d) describes the time it takes for the lowest value of RFI (lambda) to be achieved. The duration time (D, d) describes the time that lambda is maintained for, and rho (RFI/d) describes the rate of recovery of RFI until RFI = 1. There is no compensatory intake, and RFI is always < or = 1. The effects of host resistance on the values of the model parameters are proposed. Attempts were made to parameterize the model; when data were scarce, initial parameter values were derived on conceptual grounds. Predictions of the effects of pathogen dose, virulence, and host resistance are described and discussed. When comparing the responses in RFI for different genotypes, it is crucial to define the pathogen challenge (in terms of dose and virulence) and the degree of resistance of different hosts. Possible interactions between dose, virulence, and resistance were explored. Feed intake of healthy and challenged animals, at a time, may be different once the challenged animal has recovered (RFI = 1). The issue of reductions in FI during pathogen challenges is important for nutritionist and animal breeders. The large variation that has been observed for reductions in FI during pathogen challenges may be a viable point of selection. The points highlighted will aid selection strategies by quantifying the effects of pathogen dose and virulence, and time, on the FI of challenged animals. The proposed model may be integrated with other models of growth to predict animal performance during exposure to pathogens.     

PACAP对动物采食量和摄食行为的影响

垂体腺苷酸环化酶激活多肽(PACAP)于神经内分泌的肽类激素,其能抑制动物的采食,改变动物摄食行为.本文综述了PACAP的一级结构、体内分布、PACAP受体及其对动物采食和摄食行为的影响及作用机制.     

Selection response and genetic parameters for residual feed intake in Yorkshire swine

Residual feed intake (RFI) is a measure of feed efficiency defined as the difference between the observed feed intake and that predicted from the average requirements for growth and maintenance. The objective of this study was to evaluate the response in a selection experiment consisting of a line selected for low RFI and a random control line and to estimate the genetic parameters for RFI and related production and carcass traits. Beginning with random allocation of purebred Yorkshire littermates, in each generation, electronically measured ADFI, ADG, and ultrasound backfat (BF) were evaluated during a approximately 40- to approximately 115-kg of BW test period on approximately 90 boars from first parity and approximately 90 gilts from second parity sows of the low RFI line. After evaluation of first parity boars, approximately 12 boars and approximately 70 gilts from the low RFI line were selected to produce approximately 50 litters for the next generation. Approximately 30 control line litters were produced by random selection and mating. Selection was on EBV for RFI from an animal model analysis of ADFI, with on-test group and sex (fixed), pen within group and litter (random), and covariates for interactions of on- and off-test BW, on-test age, ADG, and BF with generations. The RFI explained 34% of phenotypic variation in ADFI. After 4 generations of selection, estimates of heritability for RFI, ADFI, ADG, feed efficiency (FE, which is the reciprocal of the feed conversion ratio and equals ADG/ ADFI), and ultrasound-predicted BF, LM area (LMA), and intramuscular fat (IMF) were 0.29, 0.51, 0.42, 0.17, 0.68, 0.57, and 0.28, respectively; predicted responses based on average EBV in the low RFI line were -114, -202, and -39 g/d for RFI (= 0.9 phenotypic SD), ADFI (0.9 SD), and ADG (0.4 SD), respectively, and 1.56% for FE (0.5 SD), -0.37 mm for BF (0.1 SD), 0.35 cm(2) for LMA (0.1 SD), and -0.10% for IMF (0.3 SD). Direct phenotypic comparison of the low RFI and control lines based on 92 low RFI and 76 control gilts from the second parity of generation 4 showed that selection had significantly decreased RFI by 96 g/d (P = 0.002) and ADFI by 165 g/d (P < 0.0001). The low RFI line also had 33 g/d lower ADG (P = 0.022), 1.36% greater FE (P = 0.09), and 1.99 mm less BF (P = 0.013). There was not a significant difference in LMA and other carcass traits, including subjective marbling score, despite a large observed difference in ultrasound-predicted IMF (-1.05% with P < 0.0001). In conclusion, RFI is a heritable trait, and selection for low RFI has significantly decreased the feed required for a given rate of growth and backfat.     

饲料调味剂对育成期蓝狐采食量和采食行为的影响

试验在蓝狐干粉饲料中添加肝脏香、鸡肉香等不同种类的饲料调味剂,通过对比蓝狐对基础饲粮和添加调味剂饲粮以及不同调味剂饲粮组之间的采食量和采食速度差异,确定对育成期蓝狐有诱食效果的调味剂种类以及添加剂量。饲养试验采用自由选择法试验,蓝狐自由采食对照组及试验组饲粮,试验组饲粮调味剂包括肝脏香、鸡肉香、玉米香、奶香型、甜味剂和大肠香等6种,设置了基础饲粮与每种调味剂对比,以及筛选不同调味剂之间对比总计10组对照模式。选取育成期体重相近的健康蓝狐60只,随机分成10组,每组6只,公母各半,单笼饲养。预饲期7 d,试验期14 d,每只蓝狐提供2个料盆,同时饲喂试验组和对照组饲粮,笼箱顶端安装摄像头监控蓝狐采食行为,试验期全程采集6 d的监控录像。结果表明:育成期蓝狐对不同饲粮试验组和对照组表现出不同的采食行为,奶香型与对照组偏嗜指数1.50、奶香型与鸡肉香偏嗜指数1.53,甜味剂与鸡肉香组偏嗜指数1.40,这三组的偏嗜指数高于其他组;玉米香和肝脏香组偏嗜指数均为0.99,表明玉米香和肝脏香对育成期蓝狐无明显诱食效果。采食速度最高组出现在第4组,奶香型饲量达到352.74 g/s,高于对照组和其他试验组饲粮。试验从5种饲料调味剂中筛选出奶香型和甜味剂对育成期蓝狐有显著的诱食效果,为合理利用调味剂配制适合育成期蓝狐的饲料以及提高生产性能提供理论数据支持。     

Responses in residual feed intake in lines of Large White pigs selected for growth rate on restricted feeding (measured on ad libitum individual feeding)

Residual feed intake (RFI) represents the deviation of the actual ad libitum food consumption of each animal from that predicted from combination of growth rate, backfat and metabolic body weight measurements. After 4 years of divergent selection of pigs in lines for either high or low postweaning growth rate on restricted feeding, the high line exhibited a significant reduction in RFI relative to the low line. This indicated, to some extents, a lower energy requirement for maintenance in the high than in the low line, possibly because of reduced physical activity of the animals. Estimates of genetical parameters showed that RFI was moderately heritable and those genetic correlations of RFI with carcass backfat and food conversion ratio were moderate to highly positive.     

猪饲料生物安全性相关措施综述

全球大部分养殖场都已经采用生物安全措施,以最大程度减少疾病传播过程中的病原携带者,保护猪群健康。通常认为饲料及饲料原料不会携带大量病原,但是近年来的一些意外事件表明它们可以携带并传播疾病,如猪流行性腹泻病毒(PEDV)。因此,越来越多的强调去了解饲料和饲料原料在疾病传播过程中起的作用。本文旨在综述猪饲料作为病原携带者的潜在风险,并列举了一些已经被证明可以维持猪群健康的生物安全措施。猪饲料和原料具有引发疾病的能力是综合的,包括使病原菌存活并维持病原菌在最小感染剂量以上。近期的研究表明,许多病毒(如猪流行性腹泻病毒、非洲猪瘟病毒和A型塞内卡病毒)都可以在豆粕、赖氨酸和全价料的跨境运输中存活下来。此外,PEDV在组织半数感染量感染超过5.6×101或者实时定量PCR循环阈值(qRT-PCR)为37.1时可存活下来。所有这些数据表明,存活力和传染性都可能通过饲料和饲料原料基质而保存下来。近年的研究主要集中于为防止饲料中病原菌感染猪群而采取的相关措施,包括防止病原菌进入饲料系统、热处理减轻病原菌危害、化学添加剂净化病原菌。而且,设计对策以了解病原菌在饲料生产环境中的传播,包括潜在的批次与批次之间的移动,从而能够降低传播给猪群的风险。一些饲料生物安全风险清除措施,如中链脂肪酸不仅有机会降低或清除病毒和细菌的风险,同时还能改善猪的生长性能。总之,最近几年对于饲料生物安全的关注是合理的,但是还需要进一步的研究,以充分了解饲料的风险,确定维持饲料生物安全的经济有效的措施,从而更好的保护猪群健康。     

Technical note: a monoclonal antibody-based immunoassay for determination of ractopamine in swine feeds

An ELISA was developed for routine screening of ractopamine in swine feeds. Swine feed samples were extracted and purified, and the aqueous portion of the extract was analyzed for ractopamine using ELISA and confirmed by HPLC. For swine complex feeds containing ractopamine at 2.5 to 40 mg/kg, the average recoveries ranged from 73.1 to 86.5% by ELISA and 81.9 to 98.2% by HPLC. For the swine supplement containing ractopamine at 50 to 400 mg/kg, the average recoveries were 105.5 to 111.4% by ELISA and 89.1 to 92.9% by HPLC. The limit of detection was 0.24 microg/g by ELISA and 0.48 microg/g by HPLC, respectively. Results from the swine complex feeds (P = 0.009) and the supplement (P = 0.005) using ELISA and HPLC were not highly correlated. The ELISA was more sensitive and rapid and less expensive than the HPLC method and could be used for ractopamine screening in swine feeds before confirmation and quantification by other methods, such as HPLC.     

Broiler response model to estimate the economic importance of dietary feed enzymes

The purposes of this research were: 1) a bio-assay to evaluate the effects of 2 feed enzymes (Hostazym X) and Avizyme 1505) on the performance of Cobb broilers; and 2) to calculate the economic value from the technical response data. A broiler response model that includes benefits from improved performance and incorporates the expected changes in output from Hostazym X and Avizyme 1505 is presented. Each of the 4 diets [1) Basal (corn, dried grains with solubles, and soybean meal based diet with 1,360 kcal/lb), 2) Basal+ Hostazym X, 3) Basal+ Avizyme 1505, 4) Basal+30 kcal/lb] was fed to 8 pens of 24 straight run Cobb chicks. Supplementing the basal diet with either enzyme showed significant effects on growth and feed efficiency. Quadratic equations were fitted through the response points for 49 d broilers: Body weights and feed intakes were: Basal, 6.085, 11.446; Basal+ Hostazym X, 6.231, 11.359; Basal+ Avizyme 1505, 6.165, 11.288; Basal+30 kcal, 6.455lb, 11.535 lb, respectively. Target live weights were compared at 4, 6 and 8 lb, chicken prices at $0.68 and $1.0/lb and feed costs at $200 and $400/ton. The value of feed savings by adding Hostazym X ranged from $6.92/ton for 4.0 lb broilers fed $200 feed, to $17.52/ton for 8.0 lb broilers fed $400 feed. The value of extra meat by adding Hostazym X ranged from $21.19/ton for 6.0 lb broilers selling for $0.68/lb, to $32.76/ton for 8.0 lb broilers selling for $1.0/lb. The value of feed savings by adding Avizyme 1505 ranged from $5.58/ton for 4.0 lb broilers fed $200 feed, to $12.09/ton for 8.0 lb broilers fed $400 feed. The value of extra meat from AZ1 ranged from $16.97/ton for 6.0 lb broilers selling for $0.68/lb, to $24.95/ton for 8.0 lb broilers selling for $1.0/lb. The value of the extra salable meat from feeding enzymes is 2 to 3 times greater than feed savings. Enzyme value is directly dependent on the costs of feed and meat value, and indirectly on reduced environmental costs for feed production and reduced unutilized nutrients (pollution) per lb of meat (not included in this analysis).     

Technical note: Use of near-infrared reflectance spectroscopy to predict intake and digestibility in bulls and steers

Multiple fecal samples were collected from growing Angus bulls (264 to 419 kg of BW, 3.0 to 11.4 kg/d of DMI) to predict DMI of a corn-silage-based diet. Contemporaneous digestion trials were conducted with the same diet in 12 steers in yr 1 to 3 and bulls in yr 4. Near-infrared spectra from fecal samples (n = 730 from 282 growing bulls, n = 240 from 36 steers and 12 bulls for digestion trials) were obtained from dried and ground fecal samples, and modified partial least squares regression was used to develop equations to predict DMI and DM digestibility (DMD). Although mean predicted DMI of the growing bulls (7.52 ± 0.04 kg/d or 22.4 ± 0.1 g/kg of BW) was within 2% of mean measured DMI (7.63 ± 0.06 kg/d or 22.7 ± 0.1 g/kg of BW), the mean of paired differences within samples (0.11 ± 0.04 kg/d or 0.3 ± 0.1 g/kg of BW) was greater (P < 0.01) than zero. Measured DMD (72.3 ± 0.5%) was identical (P < 0.97) to predicted DMD (72.3 ± 0.5%), and DMD for bulls in the digestion trial did not differ (P < 0.27) from DMD for steers. Prediction of intake requires incorporation of some measured values from the set of fecal samples to be predicted. Lack of similarity between spectra of fecal grab samples from the growing bulls and daily fecal collection of steers and bulls in the digestion trials in this study indicates the need for further verification before prediction of DMD with fecal grab samples.