ZEOLfTE IN PIG) DlET: EFFECT ON MLOWTH PEBFOIMANCE AND AIR QUALITP

b9

Denis Cholnière

A thesis submitted to the faculty of Graduate Studies and Research, in partlai fWfUnent of the requirements for the degree of

Master of Science

Department of Agriculturai and Biosystems Engineering McGffl Univeraie, Macdonald Campus

2 1 1 1 1, rakeshore Road Ste-Anne-de-Bellevue

Quebec, HQX W O Cenada

O Denis Choinière Febrmarg, 1090

National Library 1*1 of Canada Bibliothèque nationale du Canada

Acquisitions and Acquisilions et Bi bliographic Services senrices bibliographiques

395 Wellington Street 395. nm W.((irigton OnawaON K l A W OtbwiaON K1A MU canada Canado

The author bas granted a non- exclusive licence allowing the National Library of Canada to reproduce, loan, distribute or seU copies of this thesis in microform, paper or electronic formats.

The author retains ownership of the copyright in this thesis. Neither the thesis nor subsîantial extracts fkom it may be prïnted or otherwise reproduced without the author's permission.

L'auteur a accordé une Licence non exclusive permettant a la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/£üm, de reproduction sur papier ou sur format électronique.

L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimes ou autrement reproduits sans son autorisation.

ZeoW (77% clinoptilolite) was supplemented in grower hog rations at a rate of 2 or 6%. The growth performance (weight gain, cUîy consumption and feed conversion) and air q u Q ~ were compared against a control group's where zeoUte was replaced by fine sand. A

signifkant reduction in the feed/gadn ratio and àaiîy consumption was observed when 2% zeollte wa8 fed to pigs weighlng les8 than 40 kg. The same results were ob-ed when 6% zeolite waa fed to pi@ weigbïng more than 50 kg. No aignifïcant Merence in air quaUV (CO,, NH,, H,S arid temperature) was noticed betmeen the control and the zeollte room even if the NH, lwel feu from 12.6 to 8.7 ppm when the zeoïite level was increased from 2 to 6%. A 8-t reduction of odor intensiw was observed in the zeolite rmm, in parallel with this reaearch, a dynamic automated olfactometer for six panelists was conceived and built according to Arnerican and European guidemes. This instrument measures agricultural odors with precision and speed.

Du zéolite (77% clinoptilolite) fbt qjouté à des taux de 2 et 6%

daas la ration de porcs a l'engrais. Les performances alimentaires (gain, consommation et conversion alimentaire) et la qualité de l'air furent compar6es aux groupes témoins où le zéolite a 6té remplacé par du sable fin. Une nimlinution significative du taux de conversion menta ire et de la consommation journalière a été observée avec 2% de zéolite chez les porcs de moins de 40 kg et avec 5% de zéolite chez les porcs de plus de 50 kg. Aucune Mdrence signKicative entre la chambre contrôle et la chambre zéolite a été notée sur la quaJité de l'air (CO,, NH,, &S et température), bien qu'une baisse de 12.6 à 8.7 ppm du NH, entre l'essai à 2 et 8% fut observée âam les deux chambres. Une légère réduction de l'inteneit& d'odeur a été observée dans la chambre zéoiite. Parallèlement à cette recherche, un olfactomètre dynamique automatisé à six panelistes 8 été conçu et constlcuit selon les normes américaine et européenne. Cet appareil permet d ' m u e r de façon rapide et précise, les odeurs provenant du miiieu agricole.

First of a& 1 would iike to offer very speciel thanks to my thesis supervisor, Dr. Suzelle Bamlngton, for her guidence, support and irinumerable hours ahe devotecl& spent during my rnaeter'a work. 1 8180 want to thanlr her for the trust she demomtrated by conflàing me tihis thesis project and aJl the related work 1 did for her during m y rnaeter's.

1 wish to express my sincere appreciation to my wife Nam, who supported me during my graduate studiee, shared evergday M e and mve birth to my son Edouard and my daughter Andréa. 1 also wish to thank my parents who supported and encouraged me throughout m y studies.

1 want to thank the Fond Canadien pour l'Avancement de la Recherche and the Fédération des Producteurs de Porcs du Québec for supporting m e 5umcially; the Natural Science and Engineerfng Research Council of Canada and Shw-Gain Canada for suppomg the dynamic olfactometer project fînancially; and Nutrimin Canada for supplying the zeolite throaout the experlment.

Thanks to Denis Hatcher and al l the staff at the McGffl Swine Research Complex for the* time and help with the pig hnridling. Thanks to Dr. Bruce Downey and Dr. Chavez from the Animai Science department of McGffl University as well as Dr. Pierre Dutilleul and hi8 student Dominic Frigon from the Plant Science department for advising me during the zeolite experiment.

Thanks to Yvan Gariepy of the Department of Agricultural and Biosystems Engineering of McGill University and Mr. Donald Doyen and his assistant from the town of Montreal for their help in the conception of the dynamic ol£actometer.

Finrrlly, 1 want to thank the kind and indispensable secretaries of the Department: Sandra, Maria and Susan for helping m e out in the Mc(3i.U administrative tasks.

... lu

TABLE OF CONTENTE)

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ABSTRsCT i

RÉsTJMÉ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii TABZE OF CONTENTG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv LISTOFTABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi i LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii LIST OF SYMBOU AND ABBREVIATION8 . . . . . . . . . . . . . . . . . . . . ix CONTRIBUTION OF THE AUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . xi

CHAPTEX&I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 General Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2.1 Soil and Water Pollution . . . . . . . . . . . . . . . . 1 1.2.2 Air Pollution and Odor . . . . . . . . . . . . . . . . . 3

1 . 25 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.4 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

CHAPTERII . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - . . . . . 11 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.1 Soi1 and Water Impact . . . . . . . . . . . . . . . . . . . . . . . 11 2.2 Air QuaUty and Odor Impact . . . . . . . . . . . . . . . . . . 13

2.2.1 Sources of Air Pollution and Odor . . . . . . . 13; 2.2.2 Odor Measurement . . . . . . . . . . . . . . . . . - . 14 2.2.3 Air Poiiution Reduction and Odor Control

Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.2.3.1 Fit Manue Additives . . . . . . . . . . . 15 2.2.3.2 Feed Additive . . . . . . . . . . . . . . . . . 16 2.2.3.3 Houafng Environment . . . . . . . . . . 17 2.2.3.4 Mànure Treatment and Storage . . . 20 2.2.3.6 e SpreaàJng . . . . . . . . . . . . . . 22

2.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

iv

. . . . . . . . . . . . . . . . . . . . . . . . . . 3.8 Literature Review 32 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Objective 36

. . . . . . . . . . . . . . . . . . . . . . 8.5 Materials and Methoda 36 3.5.1 The Experimental Piggery . . . . . . . . . . . . . . 36 3.8 -2 Experimental Material . . . . . . . . . . . . . . . . 37

3.6 Resulta and Discussion . . . . . . . . . . . . . . . . . . . . . . 42 3.6.1 Animal Performance . . . . . . . . . . . . . . . . . . 42

3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 3.8Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.0 References 60 . . . . . . . . . . . . . . . . . . . . . . . . . . . C O N N E C T L N O S T . MENT 64

CHAPTER ZV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 The Design of a Versatile Dynamic Olfactometer . . . . . . . . . . 55

4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

4.2Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 4.3 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . 57

4.3.1 Ga8 Chromatography ta Meaeure Odors . . . 68 4.8.2 Electro-chexnical CeUs . . . . . . . . . . . . . . . . . 59

. . . . . . . . . . . . . . . . . . . . . 4.3.3 The Olfactometer 60 4.4 Conception Basis For a Dynamic O~ctometer . . . . 66 4.6 The Conception of The Mc(3Uï OUactometer . . . . . . . 67 4.6Summarg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 4 . 7 ~ o w l e ~ e n t . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 4.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

CHAPTERV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . GeneralConclusion 80

6.1 Recomendatione for Future Reeearch . . . . . . . . . 81 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . APPENDnA 88

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stat ist ical Mode1 88 . . . . . . . . . . . . . . . . . . . . . . . . A 1 Ekperimental Mode1 83

. . . . . . . . . . . . . . . . . . . . . . . . . . . A.2 Classical AXOVA 84

. . . . . . . . . . . . . . . . . . . . . . . . . . . 8-25 Modliled ANOVA 86 A.4'IiAANOVA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . APPENDrXB 87 . . . . . . . . . . . . . . . . . . . . . . . . . . . . OrowthPerformanceData 87

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . APPENDIXC 90 -Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . APPENDIXD 92 . . . . . . . . . . . . . . . . . . . . . . . . . . . Statiatical dnalyaia Reaults 92

LIST OF TABLES

Table 3.1 :

Table 3.2:

Table B-1

Table B-2

Table B-3

Table B-4

Table B-6

Table B-6

Table D-1

Table D-2

Characteriatics of the erperimental zeoïite . . . . . . . . . . 257

Cbaracterietics of the swine ration . . . . . . . . . . . . . . . . 30

Average Daily Gain (ADG) at 2% zeolite . . . . . . . . . . . . 87

Average Daily InWe ( D I ) at 2% zeolite . . . . . . . . . . . 87

Feed to Gain ratio (F/O) at 2% zeoïite . . . . . . . . . . . . . . 88

Average Daily Gain (ADG) at 6% zeolite . . . . . . . . . . . . 88

Average Daily Intalre ( D I ) at 8% zeoïite - . . . . . . . . . . 89

Feed to Oain ratio (F/G) at 6% zeolite . . . . . . . . . . . . . . 89

Statistical signifïc~rit levels for 2% zeoW . . . . . . . . . . 92

Statlstical signitlcant levels for 8% zeolite . . . . . . . . . . 92

vii

LIST OF FIOURES

Figure 3.1 The F/O (Feed to Gain ratio), AD1 (Average Dsily feed Intake) and ADG (Average Deily Gain) verau wewt with the feed containing 2% zeolite and 2% sand (control) (Phasel). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Figure 3.2 The FIC) (Feed ta Gain ratio), AD1 (Average Daily feed Intake) and ADO (Anrage Daily Gain) versus weight with the feed containing 6% zeolite and 6% sand <control) (Phase2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Figure 3.3 The over all carcasa index versus carcass weight for hogs fed the zeolite and sand (control feed) . . . . . . . . . . . . . . 46

Figure 4.1 The octagona1 arrangement of the system . . . . . . . . . . . 68

Figure 4.2 The airflow chart of the aystem . . . . . . . . . . . . . . . . . . . 70

Figure A-1 Experimental meut . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

Figure C-1 SAS code for the extrapolation of the missing value . . . 90

Figure C-2 S.AS code to test the nommlity of the data . . . . . . . . . . . 90

Figure C-3 SA8 code to test the hypotheses with a covariable . . . . . 90

Figure C-4 SA8 code to test the hspotfieses without a covariable . . 91

Figure C-5 SAS code to test the hypotheses by dividing the data by a covariable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

LIST OF 8YadBOLS AND ABBREVIATIONS

AD0 AD1 ANOVA A.saRAE

ASTM Ca cdn CEC

CO, COD Cu DM EN

FIG

H,S K

meq M g MANOVA DAJC N Na

=a

m 4 + NIST NSERC

P PID

RQMT SAS SD

Average Daily Gain (Wd) Average Daiïy Feed Intake (kg/d) Analyais O f V ~ c e Americen Society of Heating, Befrigerating and Air Conditioning Engïneers Arnerican Society of Testing Materiah Calcium Canadian Cation Exchange Capacity Carbon Dioxide Ghemical Oqggen Demand

C O P P ~ ~ Dry Maner Electronic Noee Feed to Gain patio Hydrogen Sulflde Potassium Miiiiequivalent Magnesiun bdultivariate Analysi8 Of Variance Montreal Urban Comrniinim Nitrogen Sodium Ammonia Ammonium National Institute of Standards and Technology Natural Science and Engineering Research Council of Canada Phosphomis Photoionization Detector Règlement ~ u r la Qualité du MLUeu du Travail (Québec) Statistical Analysis System Standard Deviatïon

LIST OF 8PadBOI.S JUID ABBREVUWïON8 (cont.)

SI Smell Intensity SiO, StUca S Z A Sodium ZeoUte A TSD-50 Signal Detection Procedure

c m a 1

introduction

1.1 General Introduction The swine industry ie the third agrlaltural aector of importance

in Canada. Z t repreaenb annuaiïy 3 billion dollars of economic activity. It also repreeents over 20 000 producere producing more tha,n 17

miiiion pigs in 1906 for which over 250% came Born Quebec. This

indus- has grown rapldly over the laat few years. For example, fkom 199 1 to 1996, the mine production has increased by 18% in Quebec, 38% in Manitoba and 8% in Canada (Statistic Canada, 1096; Stafistic Canada, 199Ta; Statistica Canada, 1997b). Even now, aman$ aii agricultural sectore, the swine industrg is the most promising.

Several provinces plan to expand the& production by 20 to 100% within the next 8 to 10 yeara (Dickson, 1996). This expansion is partially due to an increasing exportation demand. Canada exports 30% of its pork production annually to over flfty-flve Werent countries. Canada is the fiRh largeat pork exparting countzy: it exporta 836 000 metric tons of pork meat per year, primAirlïy to the United States and Asian countries (Statistic Canada, 1996; Statistic Canada, 1997a). Cana- and especially Quebec's pork, ia internationally recognbed for its high quality.

1.2 Problem Statement Internationally, the ewine indus- is Eaced Prrith strong opposition

from rural and urban commiinlties. Soil, water and air pollution caused by the swine industrg is being stated as the f;sctor reaponsible for this opposition.

1.2.1 Soil and Water Pollution Soi1 and water pollution causedm the swine indwtrg reaults fkom

poor manme storage and land disposal practices. Swine producers, in

many regions of Quebec, don% have enough land to spread theu m u r e at a rate which does not exceed the plaat's nutrienta requirementa. In the near Aiture, this situation me9 oeuse many e n ~ o n m e n t a l psoblems (Ministère de l'Environnement et de le Faune du Québec, 1996)- In fact, Quebec hae the higheat pig densi- in Cenede: a densiw of 200 pi@ versus 33 pi@ per 100 hectares of cultivated land are reported for Quebec and Canada, respectively. In some countiea of Quebec. such as in Nouvelle Beauce and Montcaïm county, thie density can reach 1800 to 1700 pi@ per 100 hectases of cuïtivated land receiving manure (Statistic Canada, lQ97a). Nonethelees, Quebec's situation is far from that of the Netherlands where thare are 378 pige per 100 hectares of rural and urban iand (Meyer, 1987).

A study conducted by the Quebec BUnWiry of the Environment and Wildllfe examined the capacity for agricultural land to recelve ariimrrl wastes (organic fertillzer) fkom Quebec fa- and it focused on the agricultural lande of Quebec's 9 largest river baeins: Chaudière, Yamaska, L'Assomption, Etchemin, Richelieu, Saint-François, Nicolet, Bayonne, and Boyer. in these basins, where there is a dense nnimni population, the agriCUItwla1 land is exceseively fertiUzed in temns of phosphorus and nitrogen. If al l the cuïtivated land of a basin could receive aU ita manure, it would be over fertillzed in phosphorus by 183% on average in all of the river basins except for that of Richelieu. The bas- would be over fertillzed in nitrogen by 186% on average in 4 (Chaudière, Etchemin, Bayonne and Boyer) of the 9 etudied basins. But manure wae found to be applied on only 29% of the cuïtivated land and on top of that, £armer8 stil l use a lot of mineral fertiïizer. The combined organic and minera1 fertiïizers applied on aU the cuïtïvated LBnd of Quebec over fertilize phosphomis by 167% and nitrogen by 133%.

O n a basin scale, the study shows that it ie possible to eee an over fertilization of up ta 460% of p h o ~ p h o ~ and 270% of nitrogen (Ministère de l'Environnement et de la Faune du Québec, 1996).

The effects of over fertillzation on the envlronment are nlfferent for phosphorus anci nitrogen. For phosphorus, over fertilization effects

are mainïy on a long-term basis. Over fertilization of phoaphorus for maqy yeam, wlll increase the level of soi1 saturation and wIU cause leaching into rivers. m e r f e m t i o n of P is responsible for an increased phosphorus level in rivers. In many large Quebec rivers, phosphorus levels exceed, several timee a year, the U t of 0.03m@l for potential eutrophicatiion (Simard et al., 1996). The effects of nitrogen over fertiïization are both short and long-term. O n a short-term basis, since it is highiy soluble, excees nitrogen is rapidly washed to rivere, but on a long-term basis, soluble nitrogen can mach and contaminate underground waters.

Underground water pollution caused by agricultural activitiee is a lurking problem for Cana- because 26 to 30% of them depend on underground water for drinking water supplg. A study conducted on 1300 domestic weïls in rural reglons of Ontario showed that around 40% of the welle contained one or more water confaminant at a lwel elrceeàing the accepîable Umit for &inking water. A correlation was found between the occurrence in weïh of bacteria, epedfically fecal coUforms, and the proximity of a farm where m u r e is routinely appïied (Betcher et aï., 1996).

1.2.2 Air Pollution and Odor The awine indus- poilutes the air by emanating e s e s auch as

methane, hydrogen sulphide and ammonia. Methane and hydrogen sulphide are dangerous for humruis and animais particdarly inside swine buil- or beside manure pita. These have been known to cause many death (Schulte, 1997). GeneraUy, methane and hydrogen sulphide me found in low concentrations and are eaeily removed. Howwer, ammonia is emïtted in a larger quantiw and because of its environmental impact, it ia more oRen stated as a pollutant.

Ammonia produced by the ewine industry represente a large amount of the total ammonia emissions. In Denmark, the agricuïtural sector is responsible for approximateu 93% of the total ammonia emiasions (Agriculture et Agroaiimentaire Canade, 1998). gay and Lee

(199'1) report that the U.K.'s agrimltural sector produces around 198 x106kg of NH, per yeaq where 28 ~10% corne ifom the mine industry. From thie 23 x106kg of NHs, 14, 7.6, 1 and 0.2 xl0'kg are released by builâlng~, land spreading, storage and outdoor pig activity, respectively. In Denmark, where the manure tanks are generally below the building, the proportions are sl-tip different : 36, 20 and 40% of ammonia emissions are pmduced by the buflding, the etorage and Born m u r e spreaàing, respectively (AgriCUIture et Agroaiimentaire Canada, 1998).

Amrnonia emissions etay in the air for a short period of time because they fa11 to the ground in dry deposits or are transfomned into other poiïutants. From 6 to 14% fkll to the ground in dry deposits directiy besides the emitting sources. In Denmark, more than 85% of these deposits will occur within 100 km of the emitting souce. The remaining 86 to 04% of the smmonia is transformed into ammonium nitrate and ammonium stilfflte when it cornes in contact with other air contaminants. Nitrate and suifate are verg smaU particles which remain in suspension in the air for a longer period of time than ammonia and can be transported up to 2 600 km away from the emitting source (Agriculture et Agpoalimentaire Canada, 1998).

Impo-t problems result from excessive ammonia emissions. They cause acid raîn that diStUrb dlfferent ecosyatems, àamage forests, a c i w n'agile ecosystems and increase the risk of river and îake eutrophications (Williams and Nigro, 199'1). Furthemore, ~mmonia emissions transformed into ammonium aerosol can, in large concentration, be hmmful ta human heaîth. in the eastern part of the Fraser Valley, in British Columbia, ammonium sulfate and ammonium nitrate constituted up to 70% of the f!ine suspended particles in the summer air which in turn reduced visibillty (Agricuiture et Agroaiimentaire Canada, 1998).

Ammonia is al80 a key ingredient in numerous odorous compounda. As a general rule, it ie said that reducing ammonia emissions by 60% ehould reduce odor by 30% (Voermans and Verdoes, 1996). Nwerthelees, odor problems are more compler than simply amnonia. In fact, odora emitted by swine production is compoeed of a mixture of more than 160 odorous cornpound8 (O'Neill and Philllpa, 1992). Liquid manue management for swine operations is by far the most uaed in Quebec and eleewhere in the world. It enhAncea me production of numerous adoraus compounds by anaerobically degrading the remlirnli of nutrients in the m u r e to, in turn, create very offensive and irritating odors.

Odor not be universally classified because perception is dependent on human emotion and memory. A research done in Southern Michigan by Lohr ( 1906) demonstrated that sociological aspects influence the perception of odors. Some correlations can be àrawn between varioua factors thet iafluence the perception of odor. For example, the demee of annoyance towards 'pig emell" is strongïy decreased when the neighborhood ha8 en economic dependance on fasming. Having lived in the area for a long time, Uving precrlously met the swine faciïiw owner or ha- the impression that the owner is making efforts to reduce the odor problem are Eactors which decrease the perception of odor as a nuisance. Moreover, the negetive perception of odor and the snnoyance with an odor is greater when the neighborhood area is categorized as residential, eub--ban or 8-

town. Those who think that odor î~ a nuisance generaLly declare that

the odor episode ie longer and more frequent than those who find that swine odor ia not a problem. Nevertheleas, aU Southern Michigan

residents sweyed during this study, decîared that they have not and do not plan to directly compiaïn about the odor problem. Nonetheless, they will support any zoning reguiation that will restrict the expansion of the swine indus- in the* region (Lohr, 1996).

Odors are hard to deal with because of their intangible nature. Very few regdations legislate odora beceuse there is no standard wrry

of measwîng them. The interaction of the different odorous compounda are so complex that it is elmast împoasible to them ~~. The human nose, with its 10 to 30 million receptor celle in 4 cma, ie still the most efacient odor sensor (id et aï-, 1997). No electronic device such as an electronic nose, &ae chromatograph and photoionizatlon detector until now, can eimulate the humas olfiwtory sense at an acceptable workable level. The ol£actometry method uses the power of the human nose ta evaïuate the concentration of odors. An olfactometer determines the threshold level for a specific odor by diluting odors below the humen threshold level and then increaaing the odor concentration unfil the odor is detected by a panelist. Henceforth, an odor unit can be defllned as the number of dilutioas required for 60%

of the population to detect the odor. Thie odor unit can be used to compare and handle dif!€erent odor problems. The olfactometer is g a i n ï n g more and more recognition around the world and is starting to become a reference method for many odor related work.

1.3 Obj ectives The dietarg inclusion of zeoW wrrs studied aa an economically

viable solution to reduce the environmental impact of the mine industry. The objectives of this atudy were to measure the effects of adding zeolite (77% cïinoptiloïite) a8 minerai supplement in the ration of grower ho@. Supplementing feed with two lwels of zeolite, 2 and 6% on drg matter basis, the following parameters were compared:

1. The average daily feed consumptson of the pigs 2. The average dailg gadn in body weight of the pi@ 3. The feed to gain ratio of the pigs 4. The carcass qiiality of the pi@ 5. The ambient room NH, and &S lwels 6. The ambient room odor intensity

In parallel with the first objective, an olfâctometer was conceived and b a t to evaluate odor concentrations in agricultural buildings. The olfactometer was not used to evaïuate the effect of zeoïite on odor levele

in piggeriea, because the instrument wsa not ready for use at the time of the trial.

1.4 Scope The resulta obtained in this etudy are iimited to two levels of

zeolite in the diet: 2 and 6%. A regular low energg and 16% protein paiiet ration waa supplemented with 2 or 6% zeolite, without balancing the energg and protein levels. Fine eilica sand was added at the same levef as zeollte in the control diet. AIRo, the 2 and 6% trial was limited to respectively 60 and 64 pige fed with the zeolite diet end respectively 60 and 54 pigs fed with the control ration. Each trial was iimited to 8 weeks. At the be#nn.ing of each triel. the pi@ had a vrrriety of weights and ages. Neverthelees, the pi@ were sorte& into two groups (zeoilte and control) of similar weight repartition with the same number of males and femaïes in each group. The results pertained ta pi@ we- from 25 to 100 kg. The etudy 1s Umited to the effects on animai powth perform~nce and ambient air q W w . The mineral mis of the feces as well as the nutrient baiance for each individual pig was not included in the study.

The automated olfactometer was buflt in accorâance to the American (ASTM) and the Eumpean (CEN) etanàard for the forced choice triangular method. The number of panellets seated was iimited to 6 , but up to 24 paneiiste can run a test in sequential mrna of 6 panelists. The dynamic dilution was compared againat NTST traceable flow calibratore. The n-butanol concentration of the butanol Lqjector was obtained by gaa chrometography. No real odor evaluation was made w i t h this olfactometer .

1.5 References

Agriculture et Agroalimentalre Canede. 1008. Stratégie de ~echerche sur la gestion du lisier de porc au Cenaàa, Calalog no. 842-7711998F. 'Ministre des Approvieionnements et Services Canada, ûttawa, Canada.

Betcher, RN., Rudolph, D.L. and Nicholson, R. 1996. Impact of agriculturl activities on groundwater quallt~~ in: Manure Management Symposium Proceeàlngs. Winllipeg, Canada, 63-62.

Dickson, A. 1998. Issues we face in developing Manitoba's livestock potentiaï- Xn: Manure Management @nnposium Proceedinga. Winnipeg, Canada, 19-26.

Kay, R.M., and Lee, P.A. 1907. Ammonia emission from pig buildings and characteristics of alurry produced by pigs offered low crude protein diets. Ln: Proceeàings of the International Symposium on Ammonia and Odour Control from Animai Production Faciiiw. Vlnkeloord. The Netherlands , 263-260.

Li, X. W., Bundy, D.S. and Hoff, S. 19S7. Dweloping an olfâctometer for ïivestock odor evahation. In: Manure Management Symposium Proceedings. Winnipeg, Canada, 166- 160.

Lohr , 1;. 1996. Factors related to odour perceptions and annoyame in a m a l context. Ln: Ma,nure Mnringement Symposium Proceedings. Winnipeg, Canada, 46-6 1.

Meyer, Th.A.M. 1997. Ammonia Control in Agriculture. in: Proceedings of the Internetional Symposium on Ammonia and Odour Control alom Animal Production Facfflty. Vinkeloord, The Netherlands, S7S-S77.

Ministère de l'Environnement et de la Faune du Qu6bec. 1996. Document de r6flexion sur la capacité des sols du territoire québécois

à supporter les élevages. UnpubUshed document of 'la table de concertation sur le projet de règlement sur la réduction de le pollution d'origine agricolen. Quebec, Canada.

O'Neill, D.H. and Phillips, V.R. 100 1. A review of the control of odour nuisance from ïivestock buildings: Part 1, Influence of the techniques for mana- waste within the bullâing. Journal of A@imltuml Engineering Reaearch, 80: 1-10.

O'Neill, D.H. and Phillips, V.R. 1892. A review of the control of odor nuisance n?om livestock builâings: Part 5, Properties of the odorous substances which have been identified in livestock wastes or in the air around them. Journal of ~ c u i t u r a l Engineering Research, 53: 23-60.

Schulte, D.D. 1007. Critical parameters for emissions. In: Proceedings of the International Symposium on Ammonia and Odour Control from Animal Production Faclïity. Vïnkeloord, The Netherlands, 23-32.

Simnrd, R.R., Cluis, D., Gangbazo, C). and Beauchemin, S. 1096. Phosphorus status of forest and agridtural soils from a waterahed of high amimal density. Journal of Environ. Qual., 24(3): 420-426.

Statistics Canada. 1996. Agriculturaï Financlal Statistics, Catalog no. 21-206-XPB. Statistic Canade, Ottawa, Canada.

Statistics Canada. 1QQ7a. AgriCU1turaï Profile of Quebec, Catalog no. QS-

176-XPB. Statistic Canada, -a, Canada.

Statistica Canada. 1997b. Faits aailïants nationaux et provinciaux du recensement de l'agriculture de 1896, Catalog no. 93FOO3S-XïF. Statistic Canada, Ottawa, Canada.

Voermans, J A M . , Verdoes, N. 1096. Reduction of ~mmonia emiRsion from pig barns. In: Proceedinga '98 of International Livestock Odor Conference '95. Iowa 8tate Uaiversiw, Amas, Iowa, Usa, 140-144.

Williams, A.& and Nigro, E . 1987. Covering slurry stores and effects on ernissions of ammonla end methane. In: Proceedlngs of the Internationaï Symposium on Ammonia and Odour Control Born Animai Production F a c i U ~ . Vinkeloord, The Netherlands, 42 1-428.

The following chapter wîU discusa different solutions ueed to reduce the environmental impacts of the BwLne indU8tfg. Amongsf all the possible solutions presently avaitable or propoeed to the producer, only thoee which are efficient and economicaQy viable are susceptible to be accepted and used, because m i n e producers are in a tdght cash situation. The following solutions will be divided into two main

categoriee: solutions to contzol or ïîmit the impact on soi1 and water, and solutions to control the sri. quailm. especielly the odor and ~mmonia emissions.

2.1 Soiï and Water impact Soi1 and -ter pollution resull3n.g from the ewine indu8try. are

maidy due to the exceseive rate of m u r e applïed to cuïtivated land. Swine m u T e N, P and K content must be reduced in order to control the environmental impact related to manue appïications. Better feed management can reduce the amount of ma,nure and can decrease its nutrient content.

Several research projects have been conducted to trg to reduce the amount of toW nïtrogen and ammonia in the fecee. Sutton et cil. (1997) fed grouiring-Plriiehing pigs with a low crude protein (10%) diet supplemented with synthetic essential amino acida and 6% cellulose to reduce the hesh manure's ammonia content by 68%. It &O reduced the menure's total nitrogen by 60% and inmeased the dry matter content of the m u r e by 60% compared to the control group fed a regular 13%

cmde protein commercial ration. Key and Lee ( 1997) showed that it is possible to reduce the volume of e l u ~ ~ g by 28% and decrease its nitrogen content by 40% by feeding a low cmde protein (16.8%) diet supplemented with synthetic essential amino acide instead of feeâing a regular commercial 22.6% crude protein diet. These studiee

demonstrated that eynthetic nmino a d supplements can reduce manure N levels : however, the cost of these synthetic axiino ad& makes thiS option les8 attractive. Lee and gag < 1987) calculated that the use of low crude protein supplements iacreaaea the operating coat by 2096, even if the reduction of feed intake and manue volume are considered.

Other growth promoters such as antibodies, hormones and beta- agonist are used to reduce the nitrogen and phosphorms excretions (Baidoo, 1996). Williams and EeQy (1994) reported that feeding ractopamine (beta-agoniet) and porcine somatdropin to a AnlRhing pig, can increase the feed etTiciencg by 0.64 and 1 .O4 kg feed(lrg gain respectively and decreaae the amount of m u r e produced by 0.68 and 1.6 1 kg of DM per finished pi$, respectively. Whlle the results prove the emciency of growth promoters, they are not accepted by Society and consumers.

The pig's feed efficiency can be enhanced by increasing the digestibility of the feedstuffs such as feeàing different sources of phosphorus. In generaî, the phosphorus suppïied by cereal grains ha8

a low digestibili~ of 20% to 40%. The rem- 60 ta 80% of the phosphorus is excreted in the feces. In contrast, the organic phosphorus in meat and bone me81 and the inorgamic phosphorus in monocalcium phosphate and dicalciuni phosphate have a higher digestibility of 70 to 80% (Jongbloed and Lenis, 1992).

The digestibility of feed can be enhanced by supplemental feed enzymes. The supplemental enzymes support the animai endogenou enzymes or supply non-existent enqmes in the digestive tract of the animai to degrade feed components. Ceilulase and phytases have been use to increase the digestibility of nitrogen and phosphorms. Jongbloed et aï. ( 199 1) ahowed that phytase, in a corn- soybean-wheat pig ration, can increase the digeatibility of the phosphomts by 36%. Furthermore, Williams and Kelly ( 1994) reported tbat phytase can reduce nitrogen and phosphorus content in m u r e by 6%. They alsa reported that cellulaee decreases nitrogen and phosphorus by S and 26-30%

respectively. Bn increase of feed effîciency can be a&ieved by microbial enzymes inclusion in a hullless b-1ey diet. Pige of 8 to 20 kg, 20 to 40 kg, 40 to 60 kg, fed with Chia diet, revealed an increase feed emciency of 10, 5.3, and 3.096, reepectively (Baidoo, 1096).

Zeolite can also be fed ss mineral supplement to increase feed efficiency, reduce ammonia voïatilization and control odors. ZeoUte is an aluminosiucate (a volcanic clay) with a high cation exchange capaclty. Some types of zeollte such as cllnoptiloiite have an af!flniw for nitrogen and sulphur compounda (Barrington and El Moueddeb, 1996). Fed to growing Aniffhing pigs at 8%. zeoïite is known to adsorb the harmful ammonla produced by the intestinal bacteria and slows down the paseage of feed to the intestinal tract. Barrington and El Moueddeb ( 1995) obtained a better net feed conversion of 0.29 kg of feed per kg of weight gain. This better feed conversion is expected to reduce the amount of manme produced and to decrease its nutrient concentration. Also, zeollte wae observed to reduce nmmonia volatïîization by 75% on average and decrease odor levels by 1 point on a scale of O to 5 (Barrington and El Moueddeb, 1996). Zeollte was also found to be interesting because its beneflts overcame its costs by 87.75 per t i h e d pig (Barrington, 1996).

2.2 Air QUtp and Odor impact

2.2.1 Sources of Air PoUution and Odor Ammonia volatiïbation is the main source of air poUution and is

closely related to odor. The rate of ammonia volatilization is a function of the dissolved ammonia in the numure and the mariure air contact area. Ammonia, emissions start right aRer the manure is excreted and continue after lt is spread on land. Ma,nrlre odors are woret wîth Uquid handling systerns resulting in anaerobic conditions. The anaerobic decomposition of swine m u r e produces many chemicelly reduced and obnoxious gases which are very offensive. This anaerobic degradetion stsrts withrn 24 hours of excretion. The concentration of malodoroue compounds increases dramatically â.om Fresh manwe to m u r e stored

anaeroblcally for 24 hours: phenol inmeases by 14096, indole by 160% and total suiphide by 1860% (O' Weff l end Philllpa, 1991). Therefore, odors are reduced by reducing the time that manue is feft inside the building. Guingand et at. (1997) ehowed that odor emiesiona were reduced by 50% when m u r e waa not mred below the slotted floor.

Control1in.g ammonln erriissione and odors can be achieved in different ways: by c o n t r o ~ the pig'e diet, controlling the building environment, applying Merent manure treatments, improving the design of manure 8tOrage tanks and flnaily, u8ing leaa odorous spreaàing methoàs. But, before evaiuating any method, an effective odor measurement device is required.

2.2.2 Odor Measurement Qwine manure odors are composed of numerous compounds such

as ca rbo~~~ l i c and phenolic acids, aldehydes, eeters, suiphide, Wols, amines and nitrogen heterocycles. O'Neffl and Phillips (1992b) have reported 168 different compounds responsible for m e odors. Over 30 of these compounàs have an odor detection thmshold at a concentration under or equal to 0.001 mg/rn3. The compounàs with the lowest detection threshold generally contained aulfur (O'Neffl and Phillips, 1992b).

Hobbs et aï. (lQQ6) compared different methods to assess odor fPom swine and poultq~ alurries. Photoionization detectors (Pm) and electronic noses (EN) based on a polypyrrole sensor were evaluated against a standard force cholce olfbctometry method. PID and EN showed some potential ta evaluate odor from swine and poultry slurries, but they are still one thousand timea lesa sensitive then the olfactometry method. Also, the results obtained with PID and EN can be different from one Earm to another slnce these device detect on& some but not al1 odorous cornpou11cb a8 an indication of the odor level. Different odorous compounds other than those detected by the device will not appear in the reeults (Hobbs et aï., 1996). OlEactametrg SU is the most accurate way to meaaure agrlcultural odors. A wider revtew

of odor measurement methode will be present in chapter IV of this document.

2.2.3 Air Poliution Reduction and Odor Control Methods

2.2.3.1 Pit Manue Additives Many manue additives are sold on the mmket. They can be

classified into 8 diffe~ent categories: ma8-g agents, counteractants, digestive deodorants, rirdeorbente and chemical deodorants.

Masking agents and counteractants treat the odors in a similnF W. Masking agents are a mixture of aromatic oils used to cover-up the undesirable odors by a more desirable one ( m e Odor Task Force, 1996). In some cases, these maaishg agents can be effective on a short- term base to control odo~s. O n a longer terrn, these compounds are quicidy broken down by the bacterial activity in lagoon8 or etorage tanks. Nevertheless, some studies ehowed that masking agents were in general more effective than digestive deodorants (Burnett and Dundero, lQ?O).

Counteractants are made-up of a mixture of aromatic oïl8 that cancel or neutrallze odors 80 that the intensity of the mixture i~ lees than that of the constituent8 (Ritter, 1989).

Digestive deodorants are made-up of bacteria and emgime that elimlnate odors through a biochemicai digestive process (Ritter. 1989). M a n y of these producte have been commerclalized during the past years. These compounds are not only sold to control odors but to enhance solid breakdown and nitrogen conservation. Under laboratory conditions, Zhu et aï. (1907) obeerved that it is possible to reduce the threshold of odore by 83 to 07%. Ritter (1989) found that digestive deodorants work when the bacteria added predominate in the m u r e . Under laboratory conditions, some producte work weU in drums of 106 or 208 liters: however, in the fleld. m u r e hancïbg conditions change and these bacteria dLe off before they become predominririt.

15

Adsorbents are products with a large surface area adsorbing the odors before they are releaaed into the environment (Bitter, 1980). Some examples of these adeorbenta are sphagnum peat moss, llmestone and zeoiite (Swine Odor Tesk Force, 1986). Airoldi et al. (1993) showed in a laboratory test that 10% zeolite (equal to 30.8 g/l of m u r e ) is needed to s#WiauWy decrease by 253% the ernission of axnmonia produced by manue. Aïlaorbent8 work effîciently only when used in large quantities and they are often tao expensive.

Chemicaï deodorant8 ~IW atrong oxWhlng agents or germicides altering or eliminating bacterial actîon reaponeible for odor production (Ritter, 1989). Oxiâiz ing agents chemïcally reduce odorous compounds into lees offensive ones (Ritter, 1989). Hyàrogen peroxide, potassium permanganate, ozone, orthodichiorobenzene chlorine formaïdehyde and paraformaidebyde have been tried to reduce odors. Hydrogen peroxide at 100- 126 ppm ha^ been found ta be most economlcal (Ritter, 1989). Zhu et aï. (1997) showed in a ïaboratory test, that some chemical orddïzing agents are able to reduce the odor threshold by 68%. Also, Berg and H6rnig (1997) added 6% lactîc ecid (60% concentrated) by volume in manue to reduce its pH to 4.5. Thia decreased the ammonia and methane emissions by nearly 90%. So in general, reducing compounds m u t be appiied fkequently in large and coetly amounts to control odors in a lagoon or Inanure tank (Ritter, 1989). Furthemore, these chernicals are often corrosive and harmAil to the environment.

2.2.3.2 Feed Additives Not all pig m u r e has the same odor production potentîal. As a

general rule, nitrogen 1s the basic ingreclient In ammonia and many odorous compounds. When the amount of protein in the diet 1s poorly balanced or protein is fed in exceas, the animal rejects this excess through its feces. Improving the pig's feed efffciency, reducea the amount of nitirogen rejected in the feces and thus reduces potential odors. In general, if aitrogen in the feces is reduced by 100 units, the odor level will be decreased by 76 unit8 ( m e Odor Ta& Force, 1898). Feed eFaciency can be imprmd in four dlnerent w: by adding

essential nmino adde. inmeasirig the digestibiliw of proteins and ad- odor absorber, enqme and microbes.

By eubetitutlng synthetlc ILmirio acids for traditional protein sources, the amount of nitrogen excreted by pige can be decreased substantally. Amino adds reduced the ammonia emissions by 40% and thus the odor emiesiona by 30%. However, these techniques are expensive and more research is needed in th18 field before they can be used commercially ( m e Odor Task Force, 1096; Lee and-, 1997).

DifFerent studies showed that it is possible to improve the digestibillw of prote- by uing a better proceesing or rende~ing technique. E m e s such as cellulases and pbgtases have been reported to reduce by 6% the amount of nitrogen in the manure (Williams and Kelly, 1994). In another study, the use of proteolytic eiilymes in feed proceasing and dietary supplements in the diet have been found to improve protein digestibility (Swine Odor Task Force, 1996).

Odor absorbent added to the pig's diet, have been evaluated. Calcium bentonite, sagebruah, charcoal and zeoîïte were used for their potential to absorb ammonïa produced in the tntestinal tract. Zeolite is particularly used because of it's high cation exchange capacity. In some cases, this cation exchange can reach 500 meq/100g (Mumpton and Fishm~n, 19'17). Aithough over 4s Werent zeolïtes are available, some such as chabasite, philUpslte, and ~Unoptilolite are more promising as feed supplement beceuse of the* epecific attraction for nmmonia (Sersale, 1983). Absorbent8 not only reduce nmmonia Ln feces, but in some cases, increase the pu's feed efEiciency. A review of the research done in this field is presented later in chapter fa.

2.2.3.3 Housing Environment By controlling dust, ammonia, and hydrogen sulflde inaide swine

housings, odor exniseions have been reduced. Different ways have been explored to control theae fbctors.

Maïodorous m e s such ammonia and hydrogen malphide as welï as reapirable dust can, at certain lwele, cause various health pmblems for pige end workers (Donham, 1989). The eafe leml of these gasee and dust is frequentïy exceeded in swine facilltiee (Donham, 1980. Donham and Popendorf, 1986. Donham, 1990). Seedorf ( 1997) reported that out of 83 livestock eheltera in Germany, nearly 58% had d u t levels exceeding health safety standards. Quebec ha8 regulations concerning the quaiity of the work environment. These regulatione set the maximum limit for ammonia, hydrogen sulphide and total duet at 26

ppm, 10 ppm and 10 mg/ma , respectively (RQaET. 1994). However. some researchers report respiratorg problems among mine workers exposed to ammania and toW dust levels of 7.8 ppm and 2.6 mg/ms (Reynolds et a., 1906). Pi@ are ale0 af!fected by the q U Q r of envtronmental conditiona. Using an electrostetic precipitator fflter. Lau et al. (1996) obtained an increase of 0.04 Irglday in daily gain by decreaeing the level of d u t by 20 to 82% in ilPi8hing hog roome. They also obsemed that the incidence of lung score was 36 to 40% lower and snout score was 26 to 40% lower in the mtered room compared to the unfiltered room. Furthermore, a laboratorg ecale test ehowed that odor threshold was decreased by 78% when 100% of the dust was removed from the exhaut air of the piggery (Hoff et al., 1997). When odors are reduced in swLne houeing facilities, the dut , ammonia and hydrogen sulphide are also reduced and workere, piga and neighbors beneflt.

A n alternative way to control dut is to use oil in the feed and on the floor. Perkins and Feddee (1996) have appïied minera1 oil to the floors of a swine fmrowing unit et a lwel of 24 mllma weekïy. They have been able to reduce d u t by 73% during a 24 hour period folïowing the application of oil. Aho, Takai et al. ( 1993) applied daily 5 to 16 ml of rapeseed oil per pig on floors of a pigïet room to reduce the reepirable dust by 76%. Zhang et al. (1994) reduced the respirable and inha,lable dust by 7696, by applylng a mineral oil to the floor of a grmer fIliishing unit. Chiba et al. (1088-1987) reduced the aeriaï dust by 21 to 83% by

ad- taïlow to the pig's ration at a level of 2.6 and 7.6%. Gore et aï. ( 1986) added 8% aoyabean oïl to a starter diet to reduce the settled dut

concentration by 46%.

A pit ventilation system can be used to control dust and air contaminante. This system should be able to decrease the total hnr?t;c?ria, ammonia and dust if the ventilation rate is high enough. In a 2 year study, conventional ventilation and plt ventilation were compared in a swine gestation building. This system was not able to reduce the bacteria at the RQMT reoornmended lm. The graham-negative bacteria were 8 to 41 tîmee hlgher than the recommended leva (Lawie et a2.. 1997). hrrthermore, Choinière et al. ( 1997) faund that pit ventilation in mshing pig units, compared to a conventional wall mount ventilation, increased ammonîa emiaeion.8 by 100% during surtuner and 20-3096 during -ter, under Quebec conditioxu.

Odor emmating f!hm a piggery can be controlled by methods such as a biofllter, a bioscrubber, a thermal incinerator, a catalytique incineration, an absorption system and m i o n chimnegs. The principle of a biofflter and bioscmibber is to paaa odorous air through a filter containing biolagical luterials (peat, compost and sou) able to breakdown volatile compounds into carbon dioxide, water and other harmless products. This method can remove 90% or more of the volatiïe organic compounda and is efacient in treating low concentrations of odorants (Swine Odor Taak Force, 1996). Hoff et aî. (1997) show that a low cost biomasa fflter maâe of chopped corn stelka and coba c m effectively remove 21 to 90% of dust particles below 5 and 10 microns respectively and reduce odor threshold by 23 to 47%. Young et al.

(1997) used a pilot-male bionlter made of a 3 to 1 mixture of yard waste compost and wood chips which reduced the odor intensitles by 61% in a swine gestation building. Dong et aL(1997) used a microbe seeded wet bioscrubber to remove up to 64% of the ammonia contained in the exhAu8t air of piggerie8. The cost of a biofllter and a bioscrmbber oRen exceed the bene-. Siemere and Van den Weghe ( 1997) concluded that the wetacrubber and biofflter were expensive and should be uaed as a laat resort.

Thermal and catalytic incineration use ternpepatures of 700 and 400 OC reepectively ta oddize odorous compounda. The absorption principle is one of the techniques that requires that air be psssed through a ffltration me- such as activated carbon (O'hfeffl et aï., 1992a). Finally, another ~ o a g ta deal with odors emanating from buildings, is to dilute the odorous compound in the atmosphere by using a welï deaigned difniaion chimney. The hemt of the c?himney is a function of the ventilation rate and the odor concentration. O'Neill et al. (1992a) estimated that a chimney 24 m in height is able to disperse the odor below ite nuisance 1-1. Chimnegs, bioscrubbers and bioîïîters can cost between 7 to 10 S pep Anished pig, and are thue the least castly methods ta control emAnAtîng OdOra from mine buildings. Therrnai and catalytic incineration and absorption can cost ikom 106 to 426 S per =shed pig and are too expenaive (O'Neffl et aï. 1992a).

2.2.3.4 Manure Treatment and Storage Manure can be treated whiïe in storage and before it is disposed

of on the land. DifTerent wags can be ueed to West manure such as composting, aeration, anaerobic digestion, aerobic digestion and biological filtration. Each method has its advantages.

Non aerated lagoons are WOW storage facilities widely used in

Southern US because they are the cheapest ways to handle maniire. Solid degradatlon i 8 promoted by the microbial activiw as long as the manue temperature is above 20 OC. Unàisturbed, the lagoon will not produce a lot of odors. But, when it's mixed and purnped, it wiï l release very offensive odors. The manure U t comes out of these lagoons is also very odorous when it ia spread on the M d .

Anaerobic digestion can produce lese offensive odors Born slurries dong with methane. Massé et d. (1997) used a laboratory scale batch -se psycbrophilic anaerobic digestor to Weat swine manitire. Wlth this treatment, COD wee reduced by up to 73% end the mnnure was relatively odorless. Peych~ophilic anaerobic digestion can produce up to 0.66 litres of methane per gram of volatile eolid fed to the digestor in

a biogas mixture containîng methane at a level of 60 to 80% of methane (Masse et d., lQQ7).

Aerobic digestion 1s abo possible to treat ewine slurry and decrease odors. The effectiveneas of the aerobic digestion is a function of residence time and alurry dry matter content. Sneath et al. ( 1992) used a farm scaïe aerobic digester ta reduce odors by 70%. A 1.6% drg matter slurry 1s aerobically treated during four dEqya to produce a stable odorless slurry for up to 30 days. Oflien, it is economical to aeparate the solid and ïiquid portions of the slurry before it îs aerated. Sneath ( 1988) showed that the higher the lm1 of dry ma- in the elurry, the cheaper it is to centriiu&ally separate and aerobically treat the mnriure, especially for large herds. Aerobic digestion will deodorize m a n u e at a cost of 1.5 to 2 S per pig, for a large herd. The solid portion ehould be composted or land apread.

m u r e tank covers reduce odor, ammonia and hydrogen sulnde production by QO%(~icul ture et Agroalîmentaire Canada, 1998; Rodd, 1998, Bundy et aï., 1997). Lee Whittington of the Prairie Swfne Center tested a balïoon-type cover to reduce odor exniasions by more than 05%.

This bdoon-type cover costs about 8 10 000 cdn for a 23 meter w e t e r concrete tank (Prairie Swine Center, 1997). Bundy et aï. ( 1997) Wied different covering materials on manure tanks and concluded that biological covers (16 to 26 cm of chopped corn sta lks or straw> are effective and inexpensive but tend to sink to the bottom of the W during the winter. A floating Leka rock layer of 4 to 8 cm offers excellent odor control all year round. It can even be partiallg reused but it is expensive ( 8 160-8 180 0.8. per ms). A low deneity poïyethylene mesh with a Uquid film on the surface does not effectively control odors. Finally, a swlEace foam produced by bubbling air through the m u r e once every three days seema to offer an interesting alternative but requires f'urther teating.

Biological fflters can integrate the treatment of m u r e and air vented from ewine buildings. First, the m u r e liquid and solid

fractions are separated. Then, the Uquid &action is treated using a biological fflter of peat moss and barh chipa. The exhaut air of the Bwine bullàing is circulated through the biological nIter as aeration sgstem. The eolid &action is compoeted or land spread (CRIQ, 1098). Such bioflltera cost approximately $10 per finished hog.

2.2.3.6 Manure Spreading Land disposal of ïivestock -tes produce a lot of odor by

distu~bing the a u r e and enhnilcing its contact with ambient air. In fact, 70% of compïaïnts concerming pig odor corne from manue spreaàing (Swine Odor Task Force, 1096). Bnother 10 % concerns the odorous air eman~ted Born the m e buil- and the remainîng 20% concerns the etorage fhciiity (Agriculture et Agroalimentatre Canada, 1998). To reduce odor exniesions and Iimmonia volatilization, manure is spread as close as possible to the ground or, even better, is directly incoqorated inta the sou. Morken and 8Bbshaug (1997) ~howed that u s i n g their new direct g~ound iqjector made it pomible to inject mnnure into the soi1 to a depth of 6 to 10 cm. This new iqjector pressurizee tne manure into a series of lamm nozzles placed àirectlg on the ground. As a result, arnmonia volatilization is decreased by up to 80% and aJl possible sources of run-off are removed.

Agriculture et Agroalimentie Canada. 1008. Stratégie de recherche sur la gestion du Usier de porc au Canaâa, Catalog no. 842-77/1998B. Ministre des Approvlaionnements et Services Canada, Ottawa, Cenada.

Airoldi, G., Balsari, P. and Chiabrando., R. 1908. Odour control in swine houes by the use of natural zeoïites: First approach ta the problem. In: Livestock Environment IV, Fourth international Symposium, ASAE. St-doseph, ddichlgan, USA, 701-708.

Baidoo, S.K. 1996. Manure modifications through dietary manipulation: Dietary effect on volume, content and odour. Zn: Manure Management Symposium Procee-. Winnipeg, Canada, 16 1 - 168.

Barrington, S.F. 1996. Practical methods for controiiing odors. In: MIinure Management Symposium Proceedin@. Winnipeg, Canada, 149- 158.

Barrington, S.F. and El Moueddeb, K. 1906. A direct method for the evaJuation of odors. Proceeding International Coaference on Livestock odors, Iowa State University. Ames, Iowa, USA, 66-09.

Berg, W. and Hornig, G. 1997. Emisaion reduction by acidiflcation of slurry - Investigations and assesament. In: Proceedlngs of the International Symposium on Ammonia and Odour Control fkom Animel Production Faciïity. Vinkeloord, The Netherlands, 469-466.

Bu-, D.S., Li, X.W. and Hoff, S.J., 1997. ~ o d o u r abaternent by covering materiala. In: Proceeàings of the International Symposium on Ammonia and Odour Control from Animal Production FaciUty. Vinkeloord, The NetheriELnds, 10 1-1 10.

Burnett, W.E. and Dundero, N.C. 19'10. Control of odors Prom animal waates. Transaction of the ASAE, 13: 22 1-224.

23

Chiba, LI., Peo, E.R. Jr., Lewis, Ad., Brumm, M.C., Fritschen, R.D. and Crenehaw, J.D. 1986. Effect of die- fkt on pi$ performance and dust levels in modifïed open front and environ en^ regulated confinement building. Journaî of Animal Science, 64: 768-78 1.

Chiba, L.I., Peo, E.R. Jr. and Lewis, A.J. 1987. Use of dietary fbt to reduce dut , aerial ammonia and bacterial colony fomning particle concentrations in awine conflnernent buildings. Transactions of the ASAE, 30: 464-468.

Choinière, Y., Marquis, B. and Gingras, 0. 1997. Ammonia and conlaminant concentrations with conventionaî veraus pit ventilation in finishing pi$ units. In: Proceedings of the Internationel Symposium on Ammonia and Odour Control from Animai Production Facîlity. Vinkeloord, The Netherlands, 365-2372.

CRIQ, 1 998. Le Biosorma-Liaier . Centre de Recherche Idustrielle du Québec, Canada. Internet document pubuhed by the CRIQ, Canada-

Dong, L., Heber, A.J., Pattereon, J.A., Strobel, B.R., Jonea, D.D. and Sutton, A.L. 1997. Bioscrmbber for removing ammonia fkorn 8wine house exhaust aîr. In: Proceedtngs of the International Symposium on Ammonia and Odour Contirol fkom Animal Production Facllim. Vinkeloord, The Netherlands, 529-682.

Do-, K.J. and Popendorf, W. J. 1986. Ambient levels of selected @ses inside swine confinement buildings. American Industrial Hygiene Association Journal, 46: 688-68 1.

DO-, K.J. 1988. Agricultuml occupational and environment heaïth: Pouce strategies for the Future. Appïied Occupational and Environment EIygiene, 4(10): F12-F22.

D o m , K.J. 1890. Health effects f!rom work in ewine corifiiiement buildings. American Journal of Industrial Medecine, 17: 17-28.

24

Gore, A.M.. Kornegrqy, E.T. and Viet, H.P. 1986. The effecta of soybean oil on nurserg air qualit9 and performance of weaning pîgs. Journal of Animal Science, 68:l-7.

Guïngand, N., Granier, R. and MARaabie, P. 1997. Characterization of air extracted from pi$ housing: effect of the presence of a1un-y and ventilation rate. In: Proceedïngs of the International Symposium on Ammonia and Odour Control fkom Anlmnl Production Facillty. Vinkeloord, The Netherian-, 49-66.

Hobbs, P.J., Misselbrook, T.H and P a , B.F. 1096. Assessrnent of odors from iîvestock waetes by a photoionization deteetor, an electronic nose, oUactometry and grrs chromatagraphy-me apectaoscopy. Journal of AgriCUItural Engineering Besearch, 60: 137-144.

Hoff, S.&, Bundy, D.S. and Li, X.W. 1997. Dust effects on odor and odor compounds. In: P r o c e e ~ of the International Symposium on Arnmonia and Odour Control fkom Animal Production Facility. Vinkeloord, The Netherlands, 10 1 - 1 10.

Jongbloed, A.W. and Lenis, N.P. 1992. Alteration of nutrition as memm to reduce environmentaï pollution by pim. Livestock Production Sciences, 8 1: 74-95.

Jongbloed, A.W., Everts, H. and Kemme, P.A. 1991. Phosphorus availabillw and requirements in pi@. In: W-Haresign and D J A . Cole (Editors), Recent Advances in Animal Nutrition. Buttemorths, London.

Kay, R.M. and Lee, P.A. 1997. Ammonia emission from pig builàings and characteristics of slurrg produced by pigs oflered low c ~ d e protein diets. In: Proceedings of the International Symposium on Ammonia and Odour Control fPom Animni Production Facfflw. Vinkeloord, The Netherlands, 268-260.

Lau, A.K., Vhcama, A.T., La, KV. and Luymes, d. 1996. Recircuïation of flltered ai^ in pig barns. Cenadîa,n AgriOU1tural Engineering, 38: 297- 304.

Lavoie, J., Marchand, O. end Gingra~, 0. 1997. Fit ventilation in pig- houeing fWiïitie8. Censdian Agrîcu la t Engineering, 3Q: 3 17-826.

Lee, P.A., and Kky, B. ad- 1997. Economic implications of reduced crude protein diete for pigs tn reduce ammonia emisslons. in: Proceeninas of the International Symposium on Ammonia and Odour Control from Anfmal Production Facility. Vinkeloord, The Netherîands, 699-706.

Massé, D.I., Droste, R.L., Kennedy, K.J., Patni, N.K. and Munroe, J.A. 1997. Potentiel for the paychrophilic aaaerobic treatment of surine m u r e ming a eequencing batch reactor. Cananinn Agricultural Engineering, 30: 26-33.

Morken, J. and Sakshaug, S. 1997. New injection technique - Direct Ground Iqfection (DGI8). In: Proceecïlnga of the international Symposium on Ammonla and Odour Control fPom Animal Production Facility. Vinkeloord, Tbe Netherlanàs , 086-690.

Mumpton, F.A. and Fishman, P.H. 1977. The appïication of natual zeoiite in animai science and aquadture. Journal of Animai Science, 45(6): 1188-1202.

O'Nelu, D.H. and Phillips, V.R. 1991. A review of the control of odour nuisance from livestock builâinga: Part 1, Influence of the techniques for manamg waste within the building. Journal of Agricultural Engineering Research, 60: 1- 10.

O'Neill, D.H., Stewart, W.I. and Phillips, V.R. 199% A review of the control of odor nuisance Prom Uvestock buildings: Part 2, The cost of odour abatement systems aa preàicted From ventilation requirements. Journal of Agricultural Engineering Research, 68: 23-60.

26

O'Neill, D.H. and Phillips, V.R. 1992b. A review of the control of odor nuisance Born Uvestock buîîdings: Part 3, Properties of the odorous substances which have been identified in liveatock -tes or in the air around them. Journal of Agricultural Engineering Research, 63: 23-60.

Perkirrs, S.L. and Feddes, J.J.R. 1996. The effect of timing of floor application of mlneral 011 on dust concentrations in a swine £arrowing unit. Canadian Agricultureil Engineering, 38: 123-127.

Prairie Swine Center, 1997. Manure lagoon cover offers odour relief. Canaàa-Saskatchewan Agriculture, Green Plan Agreement, Feature Articles. Internet document published by the Prairie Swine Center, Canada.

Reynolds, S.J., Donham, K.J., Whitten, P. ,Merchand, LA., Burmeiater, L.F. and Popendorf, W.J. 1996. Longitudinal evaluation of dose- response relationships for environmental exposurea and pulmo- f'unction in swhe producer workers. slnerican Journal of Industrial Medicine, 29: 33-40.

Ritter, W. F. 1989. Odor control of livestock wastes : State-of-the-art in North America. Journaï of Agricultural Engineering Research, 42: 51-62.

Rodd, V. 1998. The smell of manme. Nappan Research Farm, Nappan, Nova Scotia. Internet document pubbhed by the Nappaa Research Farm, Canada.

RQMT. 1994. Règlement sur la qiiAlfty du milieu du travail. 5-2.1,~. 15. Éditeur ofaciel du Quebec, Quebec, Canada.

Seedorf, J. 100'1. Human health and d u t related endotoxin concentrations in livestock builâinng In: Proceedings of the

Internationel Symposium on Ammonia and Odour Control fkom Animnl Production Faci l i~ . Vingeloord, The Netherlands, 783-740.

27

Sersale, R. 1983. Zeolite natulpali e loro utllizzazioni.La chimica e l'indwtria, 66( 12): 764-767.

Siemers, V. and Van den Weghe, H. 1997. Biofflter/wetscrubber combination for the reduction of ammonia, odour and dust exniseions of pig attening houses. In: Proceeâlngs of the International Symposium on Ammonia and Odour Control fkom Arrimni Production Fecility. Vinkeloord, The NetherlannR, 537-544.

Snesth, R.W. 1088. Centrifugation for separatîng piggery slurry 3. Economic effects on aerobic methods of odor control. JournAi of Agricultural Engineering Research, 39: 109-208.

Sneath, R.W., Burton, C.H. and WilllRmR, A.G. 1992. Continuou aerobic treatment of piggew slurry for odor conlzol scaled up to a £emn-

size unit. Journaï of Agricultural Engineering Research, 63: 8 1-92.

Sutton, A.L., Xephart, K.B., Patterson, J.A., Mumma, R., Kelly, D.T., Bogus, E., Jones, D.D. and Heber, A. 1997. Dietary manipiliAtion to reduce ammonîa and odorous compounds in excreta and anaerobic numure storage. in: Proceeâings of the Internationaï Symposium on Ammonia and Odour Control fkom Animai Production Faciïity. Vinkeloord, Tbe Netherlancb, 248-262.

Swine Odor Task Force. 1996. Option for mana- odor. Report from the North Car0Un.a Swine Odor Task Force. AgriCUItural Research Service, North Carolinn State University. (internet document, 47 pp).

Ta,kai, H., Mg$ller, F., Iversen, M., dorsal, S.E. and Bille-Hasen, V. 1993.

D u t control î n soPine builâings by spraying of rapeseed oil. In: International Livestock Environment Symposium, St-Joseph, MI, USA: ASAE. 726-738.

WUiazm, P.E.V. and KeUy, J.M. 11994. Animai production aad pollution problems. In: Uveetock Production for 81' Century. Prioritiee and research needs ed. P.A. Thacker. University of Saskatchewan Press, Saskatoon, Canada, 150-186.

Young, J.S., Clamen, J.J., Bottcher, R.W. and Westemnan, P.W. 1997. Biofiltration system for teetlng the reduction of odor from m i n e buildings. In: Proceeàhga of the International Symposium on Ammonia and Odour Control fkom Animal Production Facillty. Vinkeloord, The Netherianda, 62 1-628.

Zhaa.g, Y., Nussen, L., Barber, E.M., Feddes, J. J.R. and Sheriâan, M. 1994. Applyfng minerai ofl to reduce dust ln eurine builàlngs. Transactions of BSHRAE, 100(2): 1043-1000.

Zhu, J., Bundy, D.S., Li, X. and Rashid, N. 1997. A procedure and its application in evaluating pit additives for odor control. Canadian Agricultural Engineering, 39: 207-214,

In order to fïnd an economicaUy Piable and promising solution to the environmentaï problenis facing the mine industry, zeoïite h m been tested as a mineral supplement Lmpraving the productiviw of the m. Zeollte hmre the ab* to trap ammonla inside the intestinal tract and to increase feed efaciency. A reduction of the environmental impact is thue expected. Hence, the next chapter tests the effects of zeolite as a mineral 8upplement on m e growth performance and piggery air quaUw. The fourth chapter describes a versatile dynamic olfactometer concetved aad b a t to anaiyze the effecta of zeolite on odora exnanating fkom the experlmental rooms. The olfactometer construction was not completed et the time of the zeolite trial. Therefore, effecta of zeoW on piggery odor levels were teeted subjectively with the use of paneUats standinp dlrectly in the rooms.

Chapter three deals with the zeoïite teat. Zeoïite was used in the ration of g~owing and fhîshing pigs at levels of 2 and 6%. The e e c t on pig's growth performance and air quaiiw were teated and compared. Trials have demonstrated that it 1s possible to decrease the feed gain ratio by 0.14 to 0.19 kg of feed per kg of body weight gain with the use of 2% zeoïite for pigs of 26 to 40 kg and 6% zeoïite for pigs of 40 to 100 kg when compared to pigs receiving a ration with the sAme lwel of energy and protein.

This paper will be submitted for publication in JO- of Sioremoume Teehnolw- Authosa: Ohoiaî&se, D., B-on, 8.R and Dowaey, B. The contribution of the authors are: i) First author casried out entire experimental work with the zeolite: collected the data, anaïyzed the statistics and collaborated in the writing of the article ii) Second author supervised the reaearch and CO-eâited the article iii) Third author gave acientifîc ad-ce on the pig's nutrition and helped correct the article.

Zeolite as a mera l Supplement ta Improve Swine Productivlty and ~ s s e r g mt9

3.1 Abstract The development of the mine indu- is constrained by the

environmental impact of its manue, on water, soi1 and air. Fed as a mineral supplement, zeollte can help reduce thia impact while improving the productivitg~ of the operation. CUnoptiloUte is the preferred zeolite because of its abfflty to adsorb water and speciflcaUy NE,' in the intestine. Thus, it can slow down the passage of feed and improve the intestinal absorption of ammonium. Zeolite (77%

clinoptilolite) was fed as a minera1 supplement to a group of hogs in a specific room am3 their growth performance (feed conversion ratio, rate of weight gain, feed intake and carcase qualiw) was compared to another group of ho@ of equivalent gender aad initiai weight, houaed in a s i m i l m aaoining m m receivfng the control feed. The control feed contained fine siUca sand instead of zeolite. Two levels of zeolite were tested, 2 and 5%. The ambient air CO,, NH, and H,S lwels were monitored every week in both the control and zeolite room. The zeolite significantly improved the feed conversion rate and reduced the feed intake when supplemented at a level of 2% to ho@ weighing les8 than 40 kg. When the hogs weïghed more than 60 kg, a 5% level sigWlc8.ntly improved the feed conversion ratio and reduced the feed intake. There was no sim-iificant Merence in ambient air NH, level between the zeolite and control room, but àïiuting the feed from 2 to 6% sand or zeoïite reduced the ambient level from 12.6 to 8.7 ppm, respectively. The ventilation rate was the only factor found to affect CO, lwels and no significant amount of H,S was detected. Some 15 prrnelists directly eva,luated the air of the two rooms to find that the zeollte room had a silghtïy lower odour lwel.

3.2 Introduction Agriculture is an important ecmnomic sector for Canada. In close

cornpetition with those of cash cm- and milk, mine enterprises gross $8 billion ~mnuaUy. Furthemore, the Canadian am-food indus- has experienced in recent years, an excellent annual growth of 4.6% and employs 12% of the Canadian work force (Statistics Canada, 1997). Among al1 agricultural sectors, the swine indutry is the most promising, with several provinces planning to expand the- output by 20 to 100% withln the next 6 years. However, t-his growth program ha8 been constrained by many cornmiinities concerned about air, soi1 and water pollution.

Fed as a mineral supplement, zeoW can reduce the envfronmental impact of the swine industrg by improvlng nutrient absorption in the gut of the ho@, reducing the m u r e nutzïent content and lowering the incidence of soi1 and water pollution. By reducfng the ammonia content of manure, zeolite can lower its N volatilization.

3.3 Literature Review Zeoïite, a tektosilicate, is a volcanic mineral with a crgstalïine

hydrated aliiminosiïicate structure containing positively charged metalïic ions of the aïkaii and ailsaUne earth elements (Pewver, 1987). The crystals are characterized by SiO, tetrahedra where aU four corner orcygen ions are shared with wacent- tetràhedra to form a three-dimensional framework. Because some siïicon atoms are replaced by trivalent aliimlnum atoms, a deficiencg in positive charges arises. Thus, zeoïite possesses a cation exchange capacity (CEG) reaching in some instances 600 meq/100g as compared to Montmorfflonite with a CEC of 80 to 100 meq/100g (Mumpton and Ir'ishman, 1977). AS compared to other tektosilicates, such as feldspar and quartz, zeoîite possesses a remarkably open frnmework where void spaces can occupy up to 50% of the total volume (Airoldi et al. 1893). There exfat approximately 46 types of zeoïite with varging cavity size (molecular sieving capacity) and some zeolftes, euch as chabasite, phillipaite and cUnoptiloUte, are known to selectively adsorb NB,+ (Seraaïe, 1983).

Sodium zeoïite A ( S U ) and ciinoptilodite have been used as a mineral feed supplement because of the* respective abilltiee to adeorb Caz' and NH,' along with water. Their respective CECs are of the order of 500 and 120 meq/100g. 8W hae been produced synthetically wïth a CEC of 700 meq/lOOg. While SZA contains mostïy exchangeable Na (12.5%), clinoptilollte offers Ca and K at levels of 1.6 and 2.896,

respectively. 826 has the higheet aiTinit9 for Ca and ha8 been used to improve the adsorption of Ca and for ion exchange to reduce the toxîcity of excess salts, especially in poultry feed (Fethiere et sl., 1994; Rolland et al., 1986). Using in vitro testa, Holthaw et aï. < 1986) demomtrated that SZA can replace sodlum bicarbonate to impraw the digestibiïiQr of feed, without any negative effect on rumen Wction. Compared to a control &et, synthetic SZA st a level of 2% was found to have a negative effect on the digeatibiïi~ of dairy cattle feed (Johnson et al., 1988). whereas 2% natural SZA had no effect. In chicks, SZA waa able to exacerbate the adverse effects of excess dietaxy Ce (Watkins et al., 1989).

Clinoptilolite has a special afîiniQ~ for NH,' and has thus been used to improve nitrogen absorption in feed. The diererence in nfPLnity between S Z A and cllnoptiloîite is due to the size of the openin@ between their lattice work. Since clinoptiloïite ha8 the most potentlal for improving the growth of Uvestock whiie împroving N absorption and reducing NH, volat;illZAtion, it will be M e r examined.

Added to feed at the lwel of 6%. clinoptilolite bas improved the powth rate of domestic anrmnls and reduced manure NHs and odor emissions (Bartko et sl., 1993). Ma et al. (lB7Q) reported an increase in Utter size of 1/78 pigiets when f e e w clinoptilolite at levels of 6% to pregnant Landrace sows. However, in a second test with 6% clinoptilolite, Ma et cil. ( 1988) dld not observe any sigxW!icant effect on embrgo surviV81 and talai ovarian weight, 24 dqya aRer inseminating sows. Airoldi et sl. ( 1908) reported the *de use of zeolite by Italian farmers to reduce odor emission and iniprove feed conversion of grower hogs. However, they W e d to measure any sigrkifIcant decrease in NH,

leveb and improvement in hog Wwth when feeding 8% zeolite (88%

phillipsite, 18% chabasite, 16% bentonite and 18% illite) w i h a CEC of 265 meq 100gl. Vrzguïa and Bartko (1983) obtained an increese in weight gain of 0.48 lrghreek with ho@ fed a 6% clinoptiloilte ration as compared to those fed a regubr ration. Also, the pigs fed clinoptilolite produced leas odoriferow feces and those with nini.i.hea produced lïrmer feces within 24 hours of testing. Berrington and El Moueddeb (199s) demonstrated that 8% zeollte (77% clinoptilolite) in mine feed improved feed conversion by 0.16 kg of feed per kg of body weight gain. It aJso improved carcees quaJitq~ by 00.0Wkg and reduced manue ïüHa volatiïiz8tion during the summer.

In Cuba, weaning pi@ at 33 àays of age with an average boày weight of 6 kg were tested with different levels of cllnoptïlolite in a wheat/corn/fish diet. The best performance was obtained with 3% zeolite. Bn improved welght gain and feed efaciency were recorded as compared to pigs on a control &et con- no zeolite (Castro and Pastrana, 1980). An improved feed conversion was aïs0 obtained with the addition of 5% clinoptilolite in a mical Cuban &et for g~owing pigs from 35 to 66 kg. Thia feed contained 64% molasses. With zeolite, average daily gain was improved by 13% and the stool samples of pigs held les8 water than thoae of the control group (Castro and Elias, 1980). During the flaishing phase, from 68 to 100 kg of body weight, the best response was obtained with 7.6% cUnoptiloUte in the diet.

When used in adequate leveb, cïinoptiloïite ha8 a eignifîcant effect on the adsorption of minerals and -ter in the digestive tract of nriimals, without altering the quaïïty of the meat or product. Nestorov ( 198 1) demonstrated that cïinoptliolite in the rumen adsorbs the free NH,', helps reduce its toxïc effect. improves NH,+ ingestion by cattle and improves growth efaclency. In ruminants, protein hydrolysis alone does not suffice in providing NB,' to microbes while urea supplements generally produce excessive amounts of NH4+, the urease e m e being plentiful in the rumen. Thus, cUnoptfloiite in the presence of urea, acts as a buffer, adeorbing the exceas NH,* but releasing it for the microbes

of the munen and later in the small inteetine. TsitstsiahviU (1978) demonstrated that cilnoptiïoîite had no effect on blood when fed as a minerel eupp1ement and improved crude protein and fkee N digestibllitg fPom 73 to 76% and 88 to Q4%, reepectively for grower hom.

The CEC and leval of zeolite ueed directly etPect the performance of the animais because tao strong an adsorption effect hindere nutrient ingestion. Cllnoptrioiite with a CEC of 100 to 140 meq/100g and feed at lwels ranging between Z and 7.8% generally iniproves cattle and hog performance. When less than 1% is fed, rziriioptilolite and natural SZA have no effects (Ward et d., 199 1). Synthetic 8ZA wlth a much higher CEC negatively afFects growth, compared to the same level of natural S2.A which offers approxlmately haLt the CE0 (Elliott et el.. 109 1 ) .

3.4 Objective The objective of this research project waa to inve8tlgate the

performance of grower hogs (26 to 100 kg) fed a ration supplemented with 2 and 6% zeoïite (77% clinoptiloïite). Animai performance was measured by monitoring the rate of feed conversion, weight gaîn and feed consumption and by measuring carcass qualiw. The performance of the grower hogs fed the zeolite ration was compared against a comparable control group fed wlth a ration supplemented with fine silica sand at the same level as zeolite. The purpose of ad- sand in the control feed is to obtain tnro sîmiïar rations in terms of protein, energy and fiber content. Having simlla~ rations Uows to test the effects of zeoïite without t e s u the effects of reducing the protein, energy and fiber content due to the addition of zeoïite. Room air CO., NH, and H,S lwels were alao monttored and compared, to insure comparable conditions among experimental rooms. Furthermore, Ng,

monitoring provided information on the effect of zeolite on ambient ambient air conditions.

3.6 Materials and Methods

3.5.1 The Experimental Piggerg

The 8tudy wae conducted at the experiniental mine unit on the Macdoaald Campus of McOffl University, Ste-Anne-de-Bellevue (Montreal, Quebec) during the -ter of 1987198. This complex houses 50 SOWS in a farrow t0 W h OPeP8Uon.

The expertment w e d the two identical grower rooms measuring 14.76 m by 7.60 m, with a ceillng he-t ai3.08 m. Eachroomhas a centrai w- memuring 1.60 m in wïdth with 8 pene on each side, each of 3.00 m by 1.84 m. The nrirmAin are housed at a density of six hogs per pen or 0.766 ma/hog. Each room is ventilated by three fene, 300 mm, 400 mm and 600 mm in riinmeter, whîch are controlïed by a common thermostat. The air inlet in each room conaista of four pairs of ceillng panels, 0.9 m in length, opened agaînst a counter weight by the room negative pressure. A reclrculating duct under these slots keeps the ambient and fresh air in suapension (Agriculture Canada, 1992). The ventilation system produces a ventilation rate of 20.0 and 48.0 l/is/hog with two and three fArie in operation, respectlvel.. The pen floors are Ailly slatted and aU pige are fed ad libitum fkom standard upright feedera located at the Front of the pens.

AU hogs were weighed ueing an Ailey WeighY electronic acele wlth a capacity of 1 000 kg, an accuracy of t 0.8 kg (Wey Weighn by Weigh-Tronix, Fairmont, MN). The male indicator was a Tronix Mode1 6 15 (WeigCi-Tronix, Fairmont, MN). The CO,, NH, and &S levels in the experimental rooms where measured uaing a Multiwarn II System (Drager, Moielinger Allee, Ge-). This systern had an infra-red radiation probe to measure the CO, levela and electro-chernical sensors to measure N'Ha and H,S levels. The probes were caltbrated with standard $as concentrations, in the range of 1 and 50 ppm. The probe's accuracy wa8 equivalent to 8% of the reading.

During the Gwo phases of the experiment (November-December 1997 and Febmiarg-AprL1 1998), pi@ in grower room #2 were fed a ration supplemented with zeoîite while those in grower room #1 were given the control feed. The manue of both experimental rooms was

removed by guttar scrappers and dumped into a gravi4y flow gutter in the hall way juet outeide both lp0oms. ThFe gravîm flow gutter le& the mariiire into a prepit. Although grmer room U1 wae located closer to the manure prepit, gas traps installed at the room gutter outlet8 prevented the return of prepit m e s .

8.5.2 Experimentaï Matepial The expriment& zeoïite cantained 77% clinoptilolite and was

provided by Nutfimin Inc. It had a CEC of 130 meq/100g and a particle size mostly under 800 p m (Table 8.1).

Table 3.1 : Characteristic8 of the experimental zeolite

~ 0 p e r t S Unit Value

Particle size < 10pm 10-100- 1 0 0 - 6 0 0 ~ 6ûû-7OOpm

Chernical Properties PH 7.2 CEC mW1Wg iao

Exchangeable cations Ca meq/IOOg 30.0 Mi3 meq/ 0.63 K meq/lOOg 67.6 Na meq/lOOg 30.0

Table 3.1: charScteristic8 of the experimentzil zeoiite (cont.)

-01?ert9 Unit Value

Zeolite Content Clinoptilolite % 77 Siderite % 8 Q=- % 6 Plagîaclase % 3 K-Feldspac % 2 Barite % 3

% 2 Total % 100

Mineral Content Aliimirium rn 30.5 Barium 730 Cadmium =ma e0.028 C O P P ~ ~ 248 Magneaium mm$ a 8 8 Lead 24.3 Selenium m&rgEg e0.026 Mercurg 0.018 Potsssium mmU 6763 Ammonium Iwu 14.8 Sodium m$m 1066 Lithium m@u 3.08 Silver m&cBg -=o.as0 Srsenie m&rbg ~1.00 Wnc mlmg 31.8 cowt ~3.00 Nickel m&ckg 2.40 Chromium 6.36 Thallium m$m 0.006

Oxides Calcium Msgnesium Sodium Potassium AliImiTium Iran Mmgaaese Tit&nium Siîicon

Twice a month, the experfmental feed was manufactured by the Coopérative Fédérée MU, Ste-Rode, Quebec. It mnsisted of a standard swine ration of soybeans and corn, with 16% crude protein (Table 3.2). The zeoïite and control feed were eupplemented, respectively, with zeolite and fine silica eand. During the erperiment (November to December 1997 and Febmiarg to April 1QQ8), the energg and mude protein levels of the feed were not corrected aRer the addition of sand or zeolite.

Table 3.2: Characteristics of the swine ration

-0per4y Unit Value

Crude protein % miri. 16 Crude fat % miri. 2.0 Crude ftber % max. 5 calcuim % 0.75 Phosphorus % 0.67 Sodium % 0.20 Zinc m&(kg 140 C O P P ~ ~ m 103 Selenium mSm 0.3 Vitamin A iu/gg 9.626 Vitrrmin D-3 iu(kg 890 Vitamin Zr 26

The grower hogs used for the experiment were 76 % LannFace x 25% Yorkshire cross bred. They were raised in a weanlng room up ta a weight of 20 kg and then transferred into the grower room for the experiment. During the* L88t week in the weaning room, they were weighed and randomly assigned to one of the two groups of equal weight and sex distribution. Tmmediatelg upon being tramferreci to the grower room, the hogs were placed on the sealite or control feed.

3.5.3 Methodologg

The nutritional beneflw of zeolite as a minerai supplement in the

39

diets of pigs was evaïuated during the complete growth period of the two equal groups (same average weight, amne number of male and female) of around 80 hogs PLninhed during each two phaaes, fiom weaning lm 105 kg Uve weight. Thie represented the conditions prevaillng in a mica l commercial grower hog operation. Nevertheleas, only the reauit8 of 120 hogs and 108 ho@ in fïrst and second phase respectively has been used for the statisticaï anaJyaia since some pige were sent to the market dur- the experiment.

The experiment was splït into two phases. During the flrst phase (November-December 100'17). feed contaîning 2% of zeoîite was fed to the zeoiite groupe of pigs while feed containing 2% of sand wat3 fed to the control group of pigs. Dui?ing the second p-e. (Februarg-April1998), the level of zeoiite and sand was of 6% instead of 2%. Winter conditions prevalled during the twa phases, except for the Last week of the second phase where summer conditions prevailed for exterior temperatures reached 24°C during the day.

Animais were added to and removed fPom each experimental room on a continuai basis, rather than on an "all in aU outn basi8. O n a regular basis, groups of 40 to 60 weaned pigs were weighed, identifled and spiit into Gwa groups of identical weight and sex. One group waa tr~nsferred into one gr- m m and fed a standard swine ration with zeolite while the other group was tranaferred to the other identical grower room and fed the standard ration with aand. In each room, the weaned pigs were further eplit into aubgroups of 6 pi@ and each subgroup was placed in a single pen. Each pig was identifïed by ear notching and its pen number was recorded. Feed consumptlon and feed conversion rate were averaged for all six hogs in each pen while weight gaFn was calmlated for each individual pig.

Followlng a one week period of accilmatization to theIr environment, the experimental animais were weighed a eecond time and omcially placed on test. Feed wae weighed and added to the individual feeders once wery two àqm. The AnlmnlR were weighed evexy two

weeks, at which time the feeders from each pen were weighed to C8iculate the feed conversion rate. The ho@ were eent to the market at a weight of 100 to 110 kg. Each nnimni wae identlfied by a tattoo before being eent ta the elaughterhouse. The slaughterhouse could record, for each experimenw hog their carcess weight and index. Quebec hog producers are paid accordlng to the carcasa index which 18 determined by the carcase quality. Amongst the Werent fâctors considered to determine the carcasa index ia the carcase weight, length, color of the meat and q w t i w of fat.

The temperature inside and outside both test rooms was noted electronically everg minutes and theae recorde were used to aaus t the thermostats for an even ventilation in both experimental rooms. Every

weeks, NH,, CO, and &S levels in both roome were measured every minute over a 24 hour period using the Drager Muïtiwarn II System. The measuremenia were carried out in the center waïkwqy of the rooms, at two thirds of the length tawmds the fans, and at a height of 1,2m-

Odor evaluations were conducted once during the f i ra t stage of the elcperiment and once during the second s-ge of the experiment. Each time, a total of 16 panelista were aaked to stand in each room, and evaluate the odor lwel subjectively. The evàluatlon wae conducted in one room and then in the other. m e test took no more than 15 minutes. They were asked to rate the odor level using a scale of 1 to 6

where 1 represents a very unpronounced odor, 2 an unpronounced odor, 3 a pronounced odor, 4 a very pranounced odor, 6 a very very pronounced odor.

3.5.4 Statistical Analysis A temporal repeated-meaaurea ANOVA with a randomïzed

complete block design was used in this experiment to obseme the variation in average daily feed intalre (ADI), average daQy weimt gain (ADGC) and feed to gain ratio (WC+). The experimentel de8ign was composed of four blocks with two ta four replicates per block, twa treatments (Control and ZeolLte), four repeated-measures and one

covariable. The use of a covariable ppas necessary in thia experiment to take into account the cWEerences in weïght between groupe of pigs. The covariable corresponds to the initial weight of pigs at the beginning of each test phase. In th ia erperïment, time was the repeated fbctor (von Eden, 1993). This statisfiical model represents fktrly the trial executed, but presents some draw baclrs. Firstly, the effects of the cavariable are too strong: they overcome the expected t ime effect. Consequentïy, this model has been compared with two other 8- statistical models (1-

same model described above but without covariable, 2 - same mode1 as no. 1 but diviàing everg ADG, AD1 and F/G by the covariable) and s ï m i h r significant results are obtained. Secondly, the model does not measure the direct effect of the treatment versu8 the pig's weight. Thus by splitting the data lato two g~oups (pigs below 60 kg and pigs above 60 kg) it's possible to partially avefcome thi8 disadvantage. (See appendix A to D)

The carcass indexes were compared us- an AOV procedure where treatrnents (zeolite or sand) and block (carcass weight) effects were compared (Steel and Torrie, 1986).

3.6 Results and Discussion

3.6.1 Animal Perform~nce The performance of the hogs (ADO, AFï and FIG) is compared for

the two phases of the experiment in Figures 3.1 and 3.2. During the fïrst phase of the expertment (Figures 3.l), us- 2% zeoUte or sand, the zeolite had a eigniflcant effect on F/G and AFI for hogs w e m les8 than 40 kg. For the hogs weighlng 20 kg, the 2% zeolite feed decreased the F/G ratio by 0.14 kg, reduced the AD1 by O. 1 kg of feed, but had little effect on the ADG. For the ho@ weighlng more than 40 kg, the zeoW supplement had no sicmlficant effect. For those hogs weighing more than 60 kg, the zeolite feed had a sllghtiy depressing effect on the F/G and the AD1 witb Uttle effect on the ADG.

Ad- zeolite at a low level, to the diet of large anîmal8, had a

negative effect on their feed to gain ratio (Figure 3.1). This can be explained Born the fact that the energg and protein level of the experimental feed had not been aausted for the additive, and that the test was at-d uing anïmah of variable weiat. When 2% zeoW or sand feed was given to ho@, ali sizes of animal8 received the feed. Thus, the larger nnimniR (over 60 kg) were 8PPItched from a reguLer to a lower energg feed, w u s t e d for the additives, end 8howed eiight negative effects .

Figure 3.1 The F/G (Feed to Gain ratio), AD1 (Average Daily feed Intake) and AD0 (Average Daily m) versus wemt with the feed containing 2% zeolite and 2% sand (control) (Phase 1).

For the second phase of the experiment (Figure 3.2), the 8% zeolite feed had a slgnificant effect on the hogs we- owr 40 kg. For those hogs weighing 76 to 100 Irg, the F/Q ratio was reduced by 0.19 kg of feed per kg of 'body weight @%in. The AD1 was reduced by 0.4 kg while the ADG wa8 not affected aignificantly.

The fact that AD1 and F/G, but not the ADO, were affected by the

zeoïite indicaties that zeolite may indeed slow d m the passage of feed through the intestine of the pim. Zeoiite couid indirectly slow d o m the passage of feed by absorbing water in the intestine and reducing the AD1 of feed PPIthout airecting growth rate, thereby iniproving feed efaciency (Figure 3.2).

2 1-50

II I

1.80 j

1.00 j -A A D Q

0.50 0.60 g

28-60 50-76 76-1 O 0

Ltvs W e i g h t (Kg)

Figure 3.2 The F/G (Feed to Gain ratio), AD1 (Average Daily feed Intake) and ADG (Average Daily Oain) ver8u8 weight with the feed c o n t a i . 5% zeollte and 5% sand (control)(Phase 2) -

From this experiment, the benefits of zeolite can be expressed in ter= of a reduction of feed consumption per tinirihed pig. If a F/G decreases by 0.14 for hoga of 26 to 40 kg by using 2% zeolite and F/G decreases by 0.19 for ho@ of 40 to 100 kg by ueing 6% zeolite, thia represents a feed economy of 125.6 kg per tinished hog.

Data from the Plrst and second phase show that the energy and protein leml of the feed ha8 an effect on the feed to gain ratio. Hogs of 50 kg fed a ration àiiuted with 2% zeolite had a F/G of 3.0 while with the 5% treatment, the F/G was over 3.2.

The caxcees q W w of ell ho@ fed zeollte and eand is reported as a function of carcaes weight for all experimentel hogs (Figure 3.3). Zeolite tended to improve csrcass q W t y of ail ho@ with carcasses weighing under 88 kg although the ciifterence was not slgnifîcant due to the high standard deviation. The same reeults were obsemd by Barrington and El Moueddeb (1996). To better measure the effects of zeolite on carcam qiinlim, the erperiment needs to be repeated using 2% zeoïite when the hogs weigh lees than 40 kg, and 8% zeoiîte when the hogs weigh over 40 kg a n d u m u t be done on an 'aU in aU outn basis.

80.0 ! I

76 80 88 00 86

Caroasm W e u h t (Kg)

Figure 3.3 The over ail C B T C ~ ~ S index vereus carcasa wemt for ho@ fed the zeoïite and sand (control feed).

During the test period, no specifïc health problems were observed, except for some ta3.i biting when the hogs were inithiïy tranaferred into the grower room. Tsil biting wae more evident dufing the laat week of the second phase of the experiment when outside conditions changed rapidly from winter to summer.

3.6.2 Ambient Air QuaïiQ~ The levels of arnmonia, carbon dioxide and hydrogen suiphide are

reported in Table 8.8 for each phase of the experiment. The zeolite feed

had no aigniflca,nt effect on the levels of all three m e s , but the dilution level of the aupplement (zeolite or m d ) end the ventiïation rate had a signifïcant effect.

Table 3.3: Air quality in the experimental rooms Trial Carbon dioxide Ammonïa Hydrogen

(% by volume) ( P P ~ ) sdphide ( P P ~ )

Control Zeoiite Control Zeolite Control Zeolite

2% zeolite Nov-Dec 97 0.21 0.19 Phase 1 12-63 ~ 0 . 0 0 1 4.001 (.060)* c.040) (2.881) (8.266) Winter vent.

5% zeolite Feb-Apr 98 O. 16 O. 17 7.38 9.18 0.006 0.004 Phase 2 (.02Q) (.O371 (3.666) (3.049) (.O321 (.024) Winter vent.

5% zeolite Feb-Apr 98 0.08 0.08 3.66 Phase 2 4=21 (0.001 ~ 0 . 0 0 1 (.009) (.010) (1.299) (1.867) S u m vent.

* The values in parenthesis are the standard deviation.

At the beginning of the experiment, both experimental rooms had

an ammonia lwel of 60 ppm. Thie lwel dropped wtth the installation of gas traps inside the m u r e gutters at their outlet into the main gutter leading to the prepit. With these gae trape in place, the ambient ammonia level dropped to 12 pprn with the 2% zeolite or sand feed. Neverthelese, the embient ammonia level was lowered to an average level of 8.7 ppm when feeding 6% zeolite or sand. Thus, the feed diluted with the 5% supplement had a lower N content, than that nith the 2% supplement, and th ia had a positive effect on m u r e N volatilLzation. Also, t h i s drop cen parti- be attributed to the elightly higher ventilation rate during the aecond phase of the experiment, as opposed

to the first phase (Table 8.4). During the ïaat week of the second phaee of the experiment, the ammonia levels âropped to 3 to 4 pprn, sa the ventilation rate was drastically increa~ed. Thus, the management of the ventilation system and the leml of crude protein in the feed had a more significant effect on the ambient ammonia levels than the zeoïite supplement in the feed.

Table 3.4: Air temperature and ventilation in the experimental rooms Trial Average Fan Operation

Temperature (% of the time)

2% zeolite Nov-Dec 97 16 Phase 1 l6 92 (.408)* (.321) 92 81 30 O O Winter vent.

Feb-Apr 98 18 Phase 2 l8 96 (334) (392) 96 49 46 O O Winter vent.

Feb-Apr 98 21 Phase 2 21 99 (2.94) (2.57) 99 91 93 48 46 Sum vent.

* The values in parenthesis are the standard devïation.

The high value of the standard deviation for the NH, can be exphined by the Eact that it haa been calculated fkom several 24 hour period data and that the NH, level tende to be 3 to 4 ppm higher on average during the dey then during the night. Furthemnore, the NE, can quickly increeee by 3 to 5 ppm as soon as eomeone entera the room.

It t e e s approximateïy 48 to 60 min for thia level to return to normal.

The CO, lepel wae lower during the second p u e of the experiment because of the slightïy higher ventilation rate (Table 3.4). During the last week of the second experimental phase, it dropped to 0.08 % by volume, with the sumrner ventilation rate. Verg limlted G S levels were detected only during the second phase of the experiment.

Although the ciifference was not signiflcant, the zeoW room had a slightly higher NH, and CO, level than that of the control room. This

can be attributed to the slightly higher ventilation rate in the control room, e~pecialiy for the 4SOmm fkn (Table 25.4).

Table 3.5: Subjectdve room air odor evaluation

Trial Odor Rating ( 1-8)

2% zeolite 3.05 2.10 Nov-Dec 97 (. 844). (.626) Phase 1 Winter vent.

570 zeolite 3.16 3.16 Feb-Apr 98 (1.006) (0.8S6) Phase 2 Winter vent.

Note: the intensity of the odor rating increaaes eom 1 to 5. * The values in parentheais are the standard deviatîon.

During the first and second phases of the experiment, the level of odor in bath rooms was evaluated aubjectively. The 16 panelists evaluated the odor lwel as sllmtly lower in the zeolite room during the fhst phase and equal during the second phaae of the experiment (Table 3.5). This may have resulted Born the slightïy higher ventilation rate persisting in the control room during the aecond phaae of the experiment (Table 3.4). Again, ventilation management and room

cleaning practices mey have had a more 8ïgtWicant effect on odor level than zeolïte.

3.7 Conclusions Zeolite eupplement in the ration of Bower ho@ improved feed

efficiency when compared to feed ha- the same energg and crude protein levels. For hogs weighing les8 tha,n 40 kg, zeolite at a level of 2% reduced the feed required per kg of gain while 6% zeolïte was required ta produce eome effect for ho@ mer 40 kg. Thue, the use of zeolite at a rate increasîng frorn 2 to 6%. with the weight of the hogs, could reàuce feed coneumption by 13.6 b&(Whed hog. This repreeents a 5.8% reduction in feed consumption and probably a simiïa,r reduction in mFlriilre production. Consequently, this reduces the environmental impact of the swine production.

As compared to the control feed containing an equivalent amount of sand, zeoïite had no significant effect on the ammonia level of the ambient air in the experimental rooms. But, the dilution of the feed with 2 and 6% zeolite or sand reduced the amount of amnonia in the ambient air. Thus, zeolite can reduce the ammonia level by aUowing the use of feed with a lower crude protein lwel. The maaagement and design of the ventilation m m , the ventilation rate and the feed crmde protein level had a signifîcant effect on the ammonia and carbon dioxide. Panelists found that the odor level in the zeolite room was slightly les8 during the fkst phase of the expriment. O n a 1 to 6 scale, the zeolite zoom obtained a grade of 2 (unpronounced odor) while the control room obtained a grade of 3 (pronounced odor). Again, other factors mflv have had more infiuence on the odor level than the use of zeolite.

3.8 Acknowledgment This work wae conducted thanks to the f!inancial help of NSERC

and the Fédération des Producteurs de Porcs du Québec. The zeolïte was supplled by Nutrimrn Canada iac. of Calgary, Aîberta.

Agriculture Canada. 1902. Ventilation of smAU Uvestock. Agricuiture Canada, Ottawa, Ontario, Canada.

Airoldi, O., Balsari, P. and Chiabrando. R. 1993. Odour control in swine houses by the use of natural zeoUtea: First approach to the problem. ln: Livestock Environment IV, Fourth Intemational Symposium, ASBE, S t Joseph, Michigan, USA. 701-708.

Al-Eaaani, T., Akochi, E., MacKenzie, A.F., A U , 1. and Barringtan., S.F. 1992. Odour emission fkom liquid hog mnnure: Effect of added amendmente and aeration on odour presence and odour offeilsiveness. Journal of Enmon. QuaUty, 21 : 704708.

Barrington, S.F. and El Moueddeb, K. 1996. Zeoïite as awine manure additive to control odoura and coneerve N. CSAE paper no. 96-6 10, Saskatoon, Canada.

Bartko, H., Seidel, H. and Kovac, G. 1993. In: Zeolite '93. Fourth International Conference on the Occurrence, Properties and Utiîization of Natural Zeolites. international Commitfiee on Natural Zeolites, Department of the Earth Sciences, SU'-Colïege at Brockport, Brockport, New York 14420, USA, 43-44.

Castro, M. and Paetrana, M. 1980. Natural Zeolltes, Report Corn- de Mirieraless Tecnicos, Comintec S.A., Republica de Cuba.

Castro, M. and Elias, A. 1980. Natural ZeoUtes. Report Compania de Mineraless Tecnicos, Comintec SA., Repubiica de Cuba.

Elliott, M.A. and Edwamïs, H. M. 1991. Cornparison of the effects of synthetic and natural zeoUte on laying hens and broiler chicken performance. Poultry Science, 70: 21 15-2130.

Fethiere, R., 'Mllea, R.D. and Harms, R.H. 1994. The utilisation of sodium in Na zeollte A by broiïers. P o u l e Science, 73: 1 18- 12 1.

Holthaus, DL., Richardaon, C.R., Swift, 8-36., Gibson, A.S. and Bunger, F.L. 1996. Effect of zeolite materiela on rumen fermentation ~Wacteristlcs when compared to sodium carbonate. Journel of Animal Science, 74(supplement) : 284.

Johnson, M.A., Sweeney, T.F. and Muller, L.D. 1988. Effects of feedfng synthetic zeollte A and sodium bicarbonate on mllk production nutrient digestion and rate of digest passage in dauy COWS. Journal of Dairg Science, 7 1 : 946-9625.

Ma, C.-S., Yang, S.K., Tzeng, C.-M. and Wu., M.-K. 1983. Effect of feeâïng ciinoptflollte on embrgo sumival in m e . International CommlW on natural Zeolite, 16 1-186.

Ma, C.-S., Tzeng, C.-M., Lai, M.-K. and Teai, A.-H. 1979. The feeàingof zeolite on the Utter size at birth of ewine. Science Agriculture, 253: 203-209.

Mumpton, F.A. and Fishman, P.H. 1977. The application of natural zeolite in animai science and aquiculture. Journal of Animal Science, 45(5): 1188-1202.

Nestorov, N. 1984. Experiences wi th zeoUte and animaJ husbandry in Bulgaxia. In: Zeo agriculture: The use of natural zeoïite in agriculture and aquacutlutre. Editor8: W.& Pond and F.A. Mumpton. International Cornmittee on Natu.mil Zeoiites, Brockport, N.Y., USA, SOSpp.

Pecover, S.R. 1987. A review of the formation and geologg of natual zeoiites. In: Natural zeolite, New South Wales Geologlcal Survey, Report GS 1987/146, Department of Mineral Resources, Australia.

Pond, W. G. 1984. Reapome of growing larnbs to clinoptilolïte or zeolite Na added to corn, corn-fi& meal and cm-soybean meai die-. Journal of Animal Science, 58: 1320-1327.

Pouben, H.D. and Oksbjerg, N. 1995. Effecta of dieterg inclusion of a zeolite (clLnloptiloUte) on performance and protein metaboiïsm of young growing pi@. Animai Fee- and TechnoloBg, 83: 207-303.

Roland, D.A., Laurent, S.M. and Orloff, H.D. 1988. SheU qllAlitg a8 influenced by zeolite w l t h high ion-exchange capacity. Poulw Science, 64: 1177-1187.

Sersale, B. 1983. Zeolite naturalï e loro utillzzazion.i.Ia chimica e l'industrïa, 68(12): 764767.

Shurson, PX., Miller, E.R. and Yokoyama, M.T. 1984. Effecte of zeolite A or clinoptilolite in diets of growing mine. Journal of AnlmAl Science, 6: 1636-1646.

Statistics Canada. 1996. Agrimltural Finencial Statistics , Catalog no. 21-205-XPB. Statistic Canada, Ottawa, Canada.

Steel, R.G.D. and Torrie, J.H. 1986. Principles and procedures of statistical w i s . A biometrical approach. McOraw Hiii Publishing Corporation, New York, U . S A

Tsitsishvili, G.V. 1978. Adsorption and catalytic properties of Borne soviet natural zeoïites: in Natural ZeoUtes: Occurrence, Properties , Use. L.B. Sand and F.A. Mumpton eda., Pergamon Press, Elmsford, New York, USA, 397-401.

Vissek, W.J. 1978. The mode of growth promotion by antibiotics. Journal of Animal Science, 46: 1447.

Von Ende, C.N. 1993. Repeated-rneasure anaïysig. Growth and other time dependent meaaurea. In: Design and Analysig of Ecologlcal Experiments. Scheiner S.M. and J. Gurevitch, Editors, Cahpm~n and -11, New York, USB, 113437.

Vrzgula, L. and Bartko, P. 1988. Effects of cUnoptiloUte on weight gain and some phgsiologLca1 parameters of mine. International Cornmittee on natural Zeollte, 187-162.

Ward, T.L. ,Watkins, K.L., Southern, L.L., Hoyt, P.O. and French., D.D. 1991, interactive effects of S Z A and Cu in grawing swine: growth and bone and tissue mineral concentrations. Journal of Animal Science, 69:

726-738.

Watkins, X.L., Vagnoni, D.B. and Southern, L.L. 1089. Effect of dietary SZA and excess Ca on growth and tibia Ca and P concentration in uninfected and Eimea acervulina-infected chicks. Pouïtry Science, 68: 1236-1240.

Watkins, K.L. and Southern. L.L. 199 1. Effect of die- Na zeolite A and graded levels of Ca on gtrowth, plasma and tibia characteristics of chicks. Pouïtry Science, 70: 2296-2803.

In order to measure odors emitted by agrioultural actîvities, the next chapter will revlew technologies presently availeible to measure odors qi1Alitatively and qwtltatively. It wilf demonstrate that the olfbctometer is an instrument widely used in the rnea8urement of odor. An entirely automated and flexible àynamic olfactometer will then be preaented as it was conceived and built at the Facuit9 of Agricultural and Environmenfaal Sciences of McOill University.

This paper WU be submitted for publication in the aouxnrl of C i n a d i i n iagîa- Authon: Ohoîaî&ra, D o and Ba=rh#ton, & P o The contribution of the authors are: i) First author conceived, built and tested the olfactometer and coîlaborated in the writing of the article. ii) Second author supervised the project and CO-edited the article.

The Deeign of a Versatile Dynamic Oliactometer

4.1 Abstract Odors fkom llvestock and organic wastes are a source of

annoyame jeoparàizing many agricuitural sectors. There is a need for precise instrumentation in odor measurement to develop and test odor control techniques and reguiations. This article reviews the techniques presently available for the measurement of odors. The human noee is s t i l l the most sensitive instrument aPaileble to quan- and qualifSr odors. But, humans can introduce bias in the 8V81uation of odors.

Therefore, severaï olfactorg methods have been developed and among them, the dynamic forced cholce olEectameter is the most widely recognized. This instrument offers three sniffing ports ta a panelist, one of which is contrirninAted wlth odorous air and the panelist must iden- which port is contaminated. The first dilution presented is weli below detection or threshold ievel and the strength of the odor is incfeased until each panelist ha8 correctly identified the contaminated port twice in a row. The ASTM E6?9 method of computing the reeults is preferred and the odor concentration is presented in terms of the threshold dilution for 60% of the panelists. A panel of more than 8 and even 10 members is recomrnended for repeatable results. Before each testing session, all panel members must be rated using n-butanol.

Based on these requirements, a dynamic forced choice olfactometer was designed and b a t at McGill University. This

olfactometer is f'uîly automated for fast and accurate m i s of odor samples. The olf&ctometer ha8 a built in n-butanol unît for the rating of each panernt before every session. This n-butanol accesaerg allowa for the conversion of the dynamic ol£acfometer into an n-butanol scale olfactometer î f need be. The automation of the system and the user- frfendly inte-ce allows the controller to aaust teatingparameters with

sample strength end the confîguration of the oifbctometer with new requirements and scientiflc developmente.

4.2 Introduction Odor control is an important economic issue for ail agridtural

sectors involved in livestock production beceuse this industrg is responsible for 70% of all 0r-c waetes proàuced on a dry matter basis, including domestic wastas, pulp and paper waste8 and municipal wastewater sludge. Thia large amount of organic wastes produces important quantitlea of odors causing publlc Rnnoyance.

Among aU agricultural sectors. the mine indu- is the most promising. with several provinces planning to expand the* output by 20 to 100% witlhiii the next 6 years. Nmertheless, this growth program is met with strong opposition fkom many rural and -ban commiiriities who fear that air, soi1 and water pollution is at stake. Soi1 and water pollution can be controlled through eound manure management end spreading practices. But air pollution by odors is a more dieacult

matter to deal with because of its intangible nature and becsuse of the iimited research in this field.

For humans, odors are a subconecious stress rather than an illness. Odors affect the emotianal state but have llttle impact on the physical being of humans, althou@ these exnotional stresees can present themselves as phyaical symptorn.8 (Cunnick, 1998; Sc-, 1995). Furthemore, humane are becoming more and more sensitive to odors because of the* more Frequent expoeure to chernicale, such as food preservatives. and to drugs as for medical purposes (Sc-, 1994). Humans feel insecure and streesed when presented wlth the possibility of being exposed to odors. Thus. it ie impossible for humans to give an unbiased responee when approachad with an odor problem. Odor perceptions are subject to aii kînds of sociolo~cel bias. This bias reaults *om strese, economicai inseCuritgr and even disputes between neighbors (Thu, 1986).

Although odor control is an important issue, remarchers are just starting to understand some of the controUn.g procesees. In the fleld of odor control, the development of adequate instruments to meaaure odors ha8 been the main draw bsck.

The present article tffill review the technologies presently avaiïable to measure the qualitles of odors perceived by humans. This review will conclude that the olfactometer is the most wideïy wed instrument for the meaaurement of odors, although it offers some diaadvantages. A fully automated and flexible olfactometer will then be presented as conceived and built at the FaCUIw of Agrtcuîtu~al and Environmental Sciences of McGffl University.

4.3 Literature Revtew The @ses responsible for the emanation of odors have been

quantified and identifled well before the 1980's. It became obvious at a very early stage of research that the perception of odors by the human nose was far more complex than expected. A mixture of gases exerts a synergetic effect on the response of the human ol£actory sense.

O'Neffl and Phillips (1992) published a summeTg of al1 odorous gases identiifled by OC and associated with liveatack manue. The human olfactory sense is especially sensitive to thiols and phenolic coxnpounds with a ~ulfbr moup. The thIola are detected by the human nose at a concentration of 0 .O 1 ng/ma or at a dilution of 10'". Hydrogen sulfide is detected at a threshold of 0.10 ng/mS or et a cillution of 10'". Amrnonia is one of the hast odorous compounds, being detected at a threshold of 2 pg/ma or et a dilution of 10?

Not onïy ha8 the OC been instrumental in identiging and q u a n t m g odorous @ses emitted by livestock manure, but it has been instrumental in demonstrating that the human nose ccrn dinerentiate compounàe of verg close chemioal composition, such as uomers. For example, the human nose differentiatea the smeU of vanilla and cresol where the basis moleoular stmcture is a phenol with, for vanilla, an

OH, COH and OCHs @oup and for cresol, an O H and CHs group. The human noee detects pcresol et a concentration of 0.08 -ms, m-creeol at a concentration of 0.22 n@ms and O-cresol at a concentration of 0.4 n@m3 or et a dilution of 10-l?

The folïowing sections will review the instruments developed to quantifsr and identifg odorous @mes as well as their capablllw of representing the human olfactory reeponse.

4.3.1 Gas Chromatography to Measure Odors Gas chromatograpby (GC) is the oldeat but moat rapid and

powerful technique capable of quantifsting and i d e n w odoroue gases (Furnise et al,, 1889). It works by partitiorûng components between a mobile phase and a statïonary liquid phase retained as a surface @yer on a suitable solid supporting medium. The separation ay8tem i s contained within a column, 2 to 30 m in length and 0.60 to 4 rnrn in diameter and this column is held at a constant temperature. The choice of stationary phase for the column will be infiuenced by the polar character of the compounds to be separated.

As the human nose, the GC is capable of Werentiating between isomers and compounds with different radicals. The OC separates such compounds through the use of chiral substances or absorbent8 (March, 1992).

The OC offers specific disadmatages in odorous gas anaJysIS. It is often impossible to i d e n w ail the compounds detected (Hammond and Smith, 1981). The ClC is al80 unable to measure the synergetic effects of mixtures of gasea. Non odorous gasea such as those wi th a hydroql racïïcal, contribute to odors by chemicaï reactions or by enhaacing ionization (Hamisson et aï., 1 BQ 1 ) . Each compound ha8 its o m individual characteristics which, when mixed with other gases, form a new odor. For example, 4methyl phenol is known to increase the malodor level of a mixture and skatole is known to moâlij~ the nature of the odor (Dravnieks, 1986; Barth et a., 1984). Finally,

accurate CG) analpes are dffPlmllt to carry out in the fleld even wlth portable unit8 (gerfoot et aL, 1980)-

Direct measurement of odorous m e s by GC ia cllfficult because of the low concentrations often encountered <Schaefer, 1977). The humsn nose will recognïze some compounds at concentrations as low as 0.0 lng/m3. Thue, pre-concentration of the components of an odorous air sample is highly recommended. The volatile orga.nic compounda can be absorbed by ailica and carbon and de-sorbed using a solvent. ORen, solvent recovery is incomplete and the level of eolvent recovery must be measured before hand (Driecoiï, 1982). Extraction by Preeze vacuum techniques is an improvement cwer solvent extraction for chicken and pig slurrg odors (Yasuhara, 1987). Concentration by pur- and trapping for waste7ipater samples givee an excellent repreaentatïon of odorous components releaeed since the technique releases especially those e s e s insoluble in water (Dries and Bouguerra, 1991).

Despite the lack of insensitivity for synergetic effecta, GC and CG-MS continue to be a standard anaïytical procedure accompanying ail other technique for odor anaïpis.

4.3.2 Electro-chernical Cella and the Measurement of Odors Several sensors have been developedto detect gaseous componenta

in ambient air. These are either metal oxïde semi-conductor capacitors, chernically monrtied fleld-effect transistors, optical devices and piezo-electronic q-z cryatal devices (Sweeten, 1996). Whiie some gases can be detected by one sensor, other more complex odorant such as pyridine, requfre multiple sensors with overlapping sensitivity (MacKay-Slm, 1982).

Photo-ionization techniques have &O been used to detect air contaminanta. These are senaors where air components are ercited photo-electronicaUy and the number of excited molecules is measured. This technique 18 leas disturbed by relative humiditg than the electronic sensors but the instrument m u t be zeroed before taklng any

measurementa and thîs cause pmb1ems where no reference air is avallable. For example, to compare the odor leml of two piggeriee where air q U w may varg between roome and outside the room. a reference air aample ~lnay be nrQPicult ta obtain.

These electronic techniques offer the advantage of being portable devices, capable of memuring odors on site as -or odor wmponents are known to be cherni- unstable (Hobbs et al., 1996). The materials making up these electronic devices react with the air components and looae theIr senaitivity with tirne. Thus, they have to be repiaced regulrrrJy, adding to their coat.

Hobbs et el. < 1986) tested a photo-ionizing detector (PID) and an electronic nose (EN) a- olfactometry for the detection of pig and chicken manne odors. PID had a lin- response ratio to odor concentration (OC) and did not react to eamplee with a relative humidity exceeding 60%. Odor concentration is meaeured in terms of number of dilutione required to obtain a detection threshold by the human oIfacto~g sense. OC is defined as the number of dilutions required for 60% of -el members to detect an odor and 1s expressed in 0U/m3. PID gave a 8- down to 1 000 OU/ms but could not distinguish odore. It could detect eome compounàe down to pglm3. The EN was found to respond selectively to dlaerent types of odor components but to be lesa sensitive thnri PID. Moistue interference occurred with relative humidity above 40% and the response ratio was not linear with OC. A zero reacUng wee obtatned at 60 000 OU/mJ.

In general, electironic sensors are ideal for on site measurements but their seneitivim and selectivlw for specinc odors m e inferior to those of the human olfhctorg sense.

4.3.3 The Olfactometer The human nose is s t i l l the most sensitive instrument available

to measure odors (ASHRAE, 1993). Not only can it detect some compounde (mes01 and mol ) at levels a8 low as 0.01 ng/ma. but it can

also perceive the synergetic effects of gases in mixtures. AB a consequence, paneiiats have been uaed to measure odor fnteneity and offensiveness. The use of paneliata to measure odors W been termed olfactometry and thia technique is known to be expensive, time comiiming and exposed to bias.

Olfactometry is a complex technique requiring the proper design of several operational phases: sampling and/or bagglng the sample, carrying out the oifkctory test and computing the results (Morrieon, 1982).

Dynamic and static samplingtechniques have been used. Dynamic sampiing consists in t r~port ing the paneliste on site in an enclosed chamber, free of any odors, sampUng the source and pumping it to the oifbctometer in the wntrol -ber for immeàiate analys- by the panelists. This technique the number of panelists becauae of the iimited size of chamber being transported. It is ale0 verg costiy and the panelists can get fatigued fbom t i r aveu fkom one site to another. Static sampling requires the use of non-absorbing bags such as those of hdylarTY, TedlarTY or Tenonn (Watts et al., 1992). The Mylar bage are the least absorbing but are dif€îcult to heat se81 and any leakage from the bag can lead to bias results (Sweeten, 1998).

Odors are known to deteriorate rapiàly. Hobbs et al. (1996) obeerved that eome of the most detectable odors orddlzed aRer 2 hours of sampling. Odors collected in sampllng bags should therefore be anaiyzed withïn 8 hours of collection (O'Brien, 1908).

Once the source is sampled, the odor can be brought to the paneiists for analysis. The panellate are al- asked to rate the intensity of the odor, a quantitative measure, and also, they can be asked to evaluate the offensiveness of the odor, a q'I1Alitativ-e and more subjective measure. In oïf'actometry, the number of panel members and the rating of the panelîsts before t e s u are two maln issues affecting repeatabiliw of the results. Moet procedures recommend the use of

more than 8 paneliats (MTM, 1881; Dravniebs et al., 1886). Nice11 ( 1994) and Nice11 and Teabaloyannis (1997) rder to 10 panellsts for repeatabiliw. The rating procedure requLres the paneliata ta determine the threshold dilution of a clean air sample con- n-butanoi. Most procedures require the eiimrnRtioa of the extremely sensitive and insensitive panelista.

The olfactometry procedure can then be direct or by dilution to its threshold level. ToPd basic direct methods have been developed: one where the air sample is lpaesed through cottan or fabric awaths and the paneîiets are asked to 8meU the 8~8th~ and evaluate the internitgr of the odor ('Miner and Ucht, 1881). The other technique requiree the panelists to smell the samples themeelves, and again evaluate their odor intenefty. The rating procedure recommended for the direct waïuation of odors uses a scale of O to 10 where O peTtsins to no odors and 10 to a very intense odor (Bulley and Phillips, 1980). Barririgtan and El Moueddeb (1996) have recommended to use a scsle of O to 6 because paneliste have some diffimilw in Uerentiating between more than 6 levels. These direct methods are expoaed to a bias evaluation Born the panelista. A panelist feil to recognize a treatment reducing odor levels simplg because of an aversion for the odor ftaelf (Barrington and El Moueddeb, 1995).

To reduce bias, a butano1 male olfactometer has been designed for the direct evaïuation of eamples. This olfbctometer bas a port through which n-butanol is released in clean air at various concentrations. The undiluted odor sample 1s introduced through another 8nUEng port. The panellet is aaked to state whether or not the n-butanol concentration & stronger or weaker than that of the odor sartiple. The odor intensity of the sample is then expressed in terms of srneil intensity, SI:

S I = k C n (equation 1) where

k a n d n = 0.26 1 and 0.66 for n-butan01,

= concentration of n-butanol corresponding ta the odor aample.

Sweeten et al., (1983) tested the performance of the n-butanol scale olfactometer with 8 leveb of n-butanol (1.6 to 80 ppm), wing livestock odors. The lowest standard deviation (SD) among panellets waa obtained when the a-butanol levels were presented randomly, otherwise, the panemta tended to anticipate the result. Aho, penelfsts must be trained before using the n-butanol olfactorneter. A panel of 8 members is recommended and the stimulug flow rate of the nose piece m u t be etanderdized (ASTM, 1981). For fLeld experimenta, the n-butanol resemir must be held perfectly horizontal and at constant temperature otherwise the n-butanol concentration mag vary arnong tests.

Odor concentration is &O measured in temns of the number of dilutions required to reach the detection threshold level. The dynamic olfactometer is an apparatus ueed to measure diluted odors. It coneists of one, two or three parts, where one port releases clean air contaminated by a speciflc concentration of odor aample. The first dilution is well below the threshold value and the dilution ie reduced in sequence until the paneiîst can detect the odor. Where the panellets 18 presented with one port, he/she must answer yee or no, referring to the detection of the odor. Ta reduce biaa, the olfactometer can offer two or three ports, where only one unknown port releases the contaminated air. The panellsts m u t correctly iden* the contaminated port. This apparatus is called the dynamic forced choice olfactometer (ASTM, 1979). It is the most widely used and recognized olfactometer for the measurement of livestock and organic waste odors (Sweeten, 1998).

For both types of olfactometers, the single or multiple port increasing the atrength of the odor is achieved in equal steps. NormaUy, a panellst changes its response fPom pure guees to clear

perception in a 2 fold lncrease in cuncentzation. Ueing a step under 2 takes more time and above 3 leads to a lost of precision (Dravnieka et sl., 1986).

Once the odor ha8 been evaluated, the results need camputing into a practical form. For the n-butanol male olfactometer, the selected concentration is used to compute the SI value (equation 1). For the threshold olfactometer, the OC is presented as the l o g a r i ~ c values of the number of dilutions at the threshold, namelg the log (2) value. Three methods have been designed to compute a dilution threahold Born the response of the paneut (Dravnieka et al., 1986):

1) The ASTM E679 method giving a practical value close to that concentration for whïch p =O.6. m e value used ie a geometric mean of the concentration in 2 triangles, one in which the panelist made the la& wrong answer and the other in which the paneliet made, for the flrst m e , two consecutlve right answers.

2) The Hall-Ellis Quantal Response Metnod proposed for panels of lees than 8 members. This method mes a ranklng procedure because a log n o r m distribution of panelist sensitivim may not be satisfled.

3) The Odor Detectabiiity Function According to the Signal Detection Procedure (TSD-80) which is an index of detectebiliw, d', obtained £rom the *action of correct responses at each concentration via the use of a table.

According to Dravnieks et trl. ( l986), the TSD-60 method produces lower log (2) values than the E679 and the Hall-EliU methods for panels of more than 8 members. The E670 and Hall-EUis methoda gave similAr red t8 , but the Hall-EUS method gave leas scatter for panels of les8 than 8 members. Furthermore, the forced choice method helps reduce the panelist bias w l t h the E679 method.

Nicell et al. (1980) introduced the concept of a diecrimination

64

threshold to ïmprove result repeafabili~ by accounting for the effects of chance in 60LL1uatlng the individual panelist's threshold. This threshold is a multiple of the adual detected threshold, as computed by the E679 method but Mcluded a factor accounting for the fact that the panelist meqy still have been gueseïng during hislher last right m e r s and that the geometric mean ia an approxîmation of the true response. Nicell et al. (1886) also demonstrated the effect of selecting panelists which are either extremely sensitive or insensitive to odors. For a panel made up of 10 members, such a person can change the detection level by more t h a ~ 26%. Nicell et aï. (1886) suggeated guidelines for repeatable reeults (a variability of 10% or lese between repetitions) is to iimit the effect of one panel member on the mer aU results. A panel size of 10 members is preferable and ail panellsts with an extreme sensitivity (threshold detection level of more tban 2 dilution steps over or above that of the other panellsts) should be ellminated.

To improve the repeatabilim of results, O'Brien ( 1896) recommended asking the panelists whether or not they were guessing, &Ming or certain of the- port selection. Based on this commenta, the dilution threshold can be reported as being either that of detection or recognition.

In summarg, the most widely recognized and used technique for the meaeurement of livestock and organic waste odors is the dynamic forced choice olfactometer. The n-butanol male olfkctorneter is also used but to a more llmited extent. The dynamic olfàctometer requires that al1 panelists be rated with n-butanol before each trial. The ASTM E679 method for computing OC is al80 the most widely accepteà because of its reliabiliw and ease of application. To improve the repeatabiiity of the results, either the discrimination threshold ehould be wed or the paneliate should be asked whether or not they are gueesïng.

From this revfew, a dynamic forced-choice olfbctorneter was designed and b a t . The unît waa Ailly automated to reduce chances of errors while sett ing the dilution leveh and ta speed up the process. The

olfactometer was ale0 equippeà with a bottle of n-butanol to automatically rate 811 panellate befare each teet and to c o n . the olfactometer into an n-butanol scale olfbctorneter If need be. The following sections present the olfactometer and ite conception basis.

4.4 Conception Basis For a Dgnamic Olfactameter The dynamic oIfactometer ia the most wideîy amepted method of

evaluating the strength of an odor or the number of dilutions required before a human being can detect its preaence (Sweeten, 1998). Two recognized aseociationa have produced stanUrds on the construction requirements of olf&ctometers : sSTM (American Society of Te- of Materials, 199 1) and CEN (Comité Européen de Normaïisation, 1995). The following discussions oPill summarize the* requirements. in builàing the McGffl olfectometer, the standards of the Montréal Urban Communim (MUC) have been used, along with those of ASTM and CEN. The MUC operates an air qiIAljw laboratory dealing with odor compladnts .

A dynamic olfactometer must be so deaigned a8 to obtain an unbiased response from a paneli8t. To achieve thia, the oifbctometer must offer three air ports to a paneiist, one of which is contaminated by an odorous air sample. Thus, the paneïïst m u t try to detect the air port which is contaminated. The olfiwtometer is operated in such a way as to present the panellets with a aaanple of contaminated air at such a smai i dilution rate that the paneïist cannot detect it. The level of contaminated air is then increased by a factor of 1.4 to 2.4 until the panelist identifies the port with the contaminated air (ASTM, 109 1). Before a correct m e r is recorded, the panelm must identi$r the correct port twice in a row. The paneïist can then be asked to rate the offensivenese of the contamlnated W.

The dynamic oifactameter must be built in such a way as to avoid any foui air contaminatîon. Thus, it must be bWt of materiah which absorbing very iimited amounts of odor. The materials most recomrnended are: Teflonm and TedlarTY by Dupont, etainieas steel, glas8

and XJalophanTY by Hoechat.

The length of tubing used must be irmrted, to reduce the contamination of the system coIlstituting the dynamic olfactometer . The orifices must be as large as possible to prevent blocking. Temperature changes exceedïng 3OC mwt be prevented ta minimize airfïow discrepancies.

The àynamic olfactometer must be able to dilute air sannples from Z7 to 2 l4 with a minimum dilution range of 2" between the smallest and largest dilution capabilities. The air aample Gan be diluted in the aampiing bag by pumping a known amount of clean alr lnto the baga before sampling the contaminated air. The dynamic ol£actometer m u t al80 be able to decrease the dilution by a step 5ctor of 1.4 to 2.4 between each dilution (ASTM, 1991). GeneraUy, a step factor of 2 is used. The dilutions muet be preeented to the panellst in an aecending order, thw by incfeasing the strength of the odor.

The air porte presented to the panelista must be especially designed for snif!ibg. The airflow fkom the port m u t be higher than that of the paneïist's nose to prevent snmple dilution by the ambient air, but must present a barely perceptible fece velocity. The CEN (1998) recommends an airflow of 20 U m i n at a speed out of the air port of 0.2 to 0.5 a s . Accorw to O'Brien et 8J. (1886), the most repeatable results are obtained with air fiow ratee of 10 to 20 Umin, with fâce velodties ranghg between 0.01 and 0.1 m/s and for nose cup diametara of 4.7 and 9.6 cm. The larger noae cups tend to require a higher air flow for repeatable results. The airports muet offer a constant air distribution over tiheir entire surface, with a tolerable variation of 10%. The air ports m u t be identical as to offer no choice to the panelist.

4.5 The Conception of The McGill Olfactometer The McGffl oïfbctometer was built according to the

of ASTM (1991), CEN (1996) and the W C . It offers criteria.

epeclfications the following

The McOill oïfhctometer can seat 6 panelists et once. Acaording to the MUC, it takes 6 paneuste to obtain repeatable reBU1ts. Other researchers prefer 10 panelista (Nicelî, 1894). if ten or more panelist8 aze required for a specifïc study, the olirrctometer evaluation c a n then be repeateà using two or more sets of 6 panelists. The olfbctometer has an octagonal aeating -8ngement to minrmrzcr the length of tubing, but also to reach each evaluation panel from a central distribution point, with the same length of tubing (Figure 4.1). Each panem h a 6 three buttons, A, B and C, to eelect the contaminated port, and a siider to evaluate the offensivenees of the odor once recognized. The fcequency used to record the reeponee of the pane- is 10 Hz.

Figure 4.1 The octagonal arrangement of the system

The olfactometer distributes the air Lnto three mntn lines using precision rotameters wlth en acCU1P8Cg of 6%. The electronic mssa flow controllers used to select the range of fou1 air dilution offer an accuracy of 2%. The eirflow rate to each port is 20 Umin but it cen be mueted man- between 10 to 25 U m i n . The olfactometer can dilute the

68

contaminated sample fkom 1: 12 000 to 1:6, based on a rate of 20 U m i n at each airport. The step factor between eacb dilution can be aausted fkom 1.1 to 3. The instrument requires lesa than 6 seconds to flush the mtem and w u e t the dilution to the required lwel. The olfactometer offers three dinerent dehya aus t ab l e fkom O to 6 minutes: ûrst, a delay between each dilution below the threshold level; second, a delay between each dilutlon above the Weshold level; Chird, a delay between each air sample. B8TM (1991) recommends from 2 to 6

minutes between a m operations when panelista are expoaed to a strong odor, to minimize their £stigue. Allowing 0.6 minutes between sni.fRnP operations under the threehold and 2 minutes above threshold, and if 8 odor lmIs muat be tested before exceedlng the threshold level, the present olEactameter can run one 8ample in lees than 10 minutes-

\

The air flow chart of the system ie fflustrated in Figure 4.2. The olfactometer ha^ a built in n-butanol unit alla- for the automatic calibration of aU panelista before each test. The n-butanol concentration et the outlet of the iqjector ie 3 480 puma. The olfactometer is conceived to callbrate the panelista uising a range of n-butanol of 0.36 to 345 pI4rnS.

The olfactometer can hFllldle four air sample bags et once. The air bags are placed in a stainless steel t i and compressed to force the contaminated air into the system. The computerized sy8tem of the olfactometer identifles the bag by the outlet it is connected to.

The entire system is fully automated. Before any test, the supemiaor sets hi8 requirementa fkom the computer screen. Then, the panelists are seated and the computer of the system runs the entire test. The computer initlally tests aU prrnebts using n-butanol, if it is required by the eupervisor. The computer stops when all panelista have correctly identifled the correct n-butanol airport twice in a row. During the n-butanol evaluation, the computer is programmed to mave on to the next higher dilution once the paneliate have 811 answered. The contsminated port ia randomly changed by the computer between each

dilution. Then, the cornputier proceeds to the evaluation of the first air sample and proceem with the detendnation of the detection fhreahold as for the n-butenol test. In turn, the computer wFll test the other three air samples. The panellets can record the offensiwness of the air sample whenever they want and the computer will autnmetically record their observation. A suder ia uaed for this purpose. Once the evaluation is completed, the computer prints out the resulfs. Each panelist has a digital indicator where instructions are dieplayed by the computer.

To 6 panels

120 L/mLn

Air Distributor &- Air Distributor TO 6 pn8b TO 6 panels

Figure 4.2 The airflow chart of the m m

The computerized system is set up to record the name of the paneïist and their individual n-butanol rating from one test to another. The computer system can contra1 aU tasks, such as the pump airflow, the adjustment of the settings, the flow direction, the butanol caiibration and the flushing and purging operations. It recorde aU data, such as the origln of the contaminated air sample, the time and date of

sampUng and ita evaîuation. AU information ia produced on screen or can be printed out. Thia automation a ï l m for the epaluation of 4 contaminated eir samples with 6 panellete, within 40 minutes and depending upon the epeed of the panelista. The aystem is al80 set up to compute the results according ta the E670 method or the Hall-Ellis method.

Finaîïy, the airflow was verifled against a flow calibrator model Dry-Calm DC-1 with flow cella model Drg-CalTY DC-1L for a fiow below 1 Umin and DC-1H for a flow above 5 U m i n (Dry-CalTY by BI08 International Corp., Pomptom Plaine, N.J., USA). From thi8

calibration, the maximum variation in air flow ha8 been found to be 2% of the set flow for aU the ma88 flow controllers. Furthemore, the flow emaaating fkom 8- fi.m.neIs have been found to be preciee within 2% of the set flow rate. Consequently, the maxium caiculated error of a àiïution at the 8nIfRng port 18 4%.

4.6 S u m m a r y

The McGffl ol£actometer is a dynamic forced choice ol£actometer M y automated which is capable of caltbrating the panelists and

4 contaminated air samples within 40 minutes. It seats 6 panelists but the ayetem can be built to accommodate more panelists if need be. Most researchers prefer to use 10 panelists for repeatable results. This olfactometer is unique because of its lwel of automation and apeed with which it can evaîuate air samplee. It requires the controller to input the base parameters, the nnme of the paneliste and the sample origla. Once the panelista are seated, the computer muis the test without any M e r input from the controlier. The results are anaiyzed using either the E670 method or the FTnll-EUS method and are printed or saved into the database. The n-butanol rating of each paneïist is ale0 memorized as hture reference on the sensitivity of each person.

The M c W University olfactometer can be used as an n-butanol male olfactometer if required, because it haa a built in system capable

of producing n-butanol dilutions. The only limitation is the presentation of 20 Umin of undiluted conîaminated 88131p1e. if such a sample must be preaented for 1.6 minutes to 6 paneîlsta, the sample volume m u t exceed 200 1, which repreeenta the content of the four sample baga which the sgstem can hold.

4.7 Acknowleàgment The McOffl olfsctometer wss built with the coliaboratlon of the

IMUC, NSERC aind S ~ U ~ - G & I .

ASHRAE. 1993. Handbook of funcbmentals. Ameaican Sociew of Heating, Refrigeration and Bir Conditioning. Mianta, Oeorgia, U.S.A., 12.1-12.6.

ASTM 1 B 8 1. Standard recommended practices for referencing supra-threshold odor intensity. dhericsn Society for Tes- and Materiais, Philadelphia, PA, USA.

ASTM 1991. Standard practice for determination of odor and -te thresholds by a forced choice ascending concentration seriee method of limits. 3679-9 1. 198 1 Annual Book of Standards, hnerican Society for Tes- and Materials, Phlladelphla, PA, USA.

ASTM. 1979. Determination of odor and teste thresholds by forced choice ascenàing concentration series method of iimits. E 6'19-79. Americaa Society for Tes- Materiala, Phihdelphia, PA, USA.

Barrington, S.F. and El Moueddeb, K. 1996. A direct method for the evaïuation of odors. Proceeàing Internationaï Conference on Llvestock odors, Iowa State Universim, Ames, Iowa, USA, 66-69.

Barth, CICL., Eliïott, L.F. and Melvln, S.W. 1984. Using odor control technology to support animai agriCU1ture. Transactions of the ASAE, 27(3): 869-864.

BUlley, N.B. and P W p s , D. 1980. Sensorg evaluation of agricultural odors. A critical review. Canadian Agricultural Engineering,, 22(2): 107-1 12.

CEN. 1996. Odor Standards. CEN/TC 264N 134. Comité Européen de Normalisation, Dusseldorf, Gemnany.

Cunnick, J. 1996. Implications of environmental odor on peychologîcal statua and heaïth. In: International Symposium on Uvestock Odor Control. Iowa State University, -es, Iowa, USA, 186-189-

Dravnieks, A. 1886. Atlas of odor characMr proflies. BSTM Committee on sensory evaluation of materials and products. ASTM data series, Baltimore, MD, USA.

Dravnieks, A., Schmidstsdorff, W. and Meilgaard, M. 1986. Odor threshold by forced choice dynamic triangle oifactometer: reproducibïiity and metho- of calCU18tion. Journal of the A ~ P Pollution Control Association, 36: 900-006.

Driscoll, J.N. 1982. Identification of hydrocarbons in complex mixtures using a variable energg PID and capilhq column gas chromatograph. Journal of Cbromstographic Science, 20: 9 1-94.

Driss, M.R. and Bouguerra, M.L. 1991. Analysis of volatile organic compounds in water by purge-trap and gas chromatography. Journal of Environmental -cal Chemistry, 48: 198-204.

Furniss, B.S., Hannaford, A.J., Smith, P.W.G. and TatcheU, A.R. 1989.

Vogel's textbook of practicaï organk chemistry. Fïfth edition. Longman House, Burnt Mill, Harlow, Essex, U.K.

Hammond, E.G and. Smith, R.J. 1981. Survey of some molecular dispersed odor constîtuents in mine house air. Iowa State Journal of Research, 66(4): 393-309.

Hardwick, D. C. 1986. Agridtural problems related to odor prevention and control. in : Odor prwention and control of organic aludge and livestock f e r a . Elsier Applied Science Publi~hers , London, England, 21-26.

Harrison, R.M., de Mora, S.J., Rapsomanikis, S. and Johnson, W.R. 199 1. ïntroductorg Che- for the environmental sciences. Cambridge United Press, Cambridge, U.K., 272-275.

Hendrick, J., Vrielink, M. and van der Peet., C. 1997. Reducing ammonia emissions From pig housing by addng acid salts to the feed. International Symposium on Livestmk Environment. American Society of AgrimltumJ Engineering, St Joseph, Michigan, USA.

Hobbs, P.J., Misselbrook, T.H. and Pain, B.F. 1996. Assessrnent of odors fkom livestock wastes by a photoionWatton detector, an electronic nose, olfactometrg and gee &matograPhy-ma68 spectroecopy. Journel of Agricuïtural Engineering Research, 60: 137-144.

Kerfoot, H.B., Pierret, S.L., Anilr, E.N., BottreU, J.V. and Petty, J.D. 1990. LbalyUcal performance of four pomble 0.C. under fleld conditions. Journal of Air Quality and Waste Mamagement Ass., 1106-1 114.

MacKap-Sini, A. 1992. Electronic odor detection - problems and possibilities. In: Odor update '92, Proceeàings Worlzshop on Agricultural Odors. MRC report Department of Agriculture of Queensland # 64/24. Department of Primary industries, Toowoomba, Queensland.

March, J. 1992. Advanced organic chemistry. Reactions, mechanics and structure. Fourth edition, John Wiley and Som, New York, USA.

Miner, J.R. and Licht, LA. lQ8 1. Fabric m tches as an aid in îivestock odor evaluations. In: Livestock -te, a renewable resource. Procee- of the 1980 Internationaï Symposium on Animal Waste. American Society of Agrlcultural Engineers, St Joseph, Michigan, USA, 302-305.

Morrison, G.R. 1982. Meaeuement of flavor threshold. Journal of Inst. Brew, 88: 170.

Nakamoto, T, Fudkuda, A. and Morizumi, T. 1991. ïmprwement of identification capabillQr in an odor sensing system. Sensors and actuators, B3: 221-226.

Niceii, J.A. and Tsakaloyanni8, M. 180'7. A protocol for odor impact assessment. ProceecUng8 of the 4th Internationaï Conference on Characterization and Confrol of Emissions of Odors and VOCs, Air and Waste management Association. Pittsburg, Penxmylwmia. October 20-22, Montréal, Canada, 182-194.

Nicell, J. 1994. Development of the odor impact model as a regulatory strategy. Internationai Journal of Environment and Poilution, 4 (%a):

124138.

Nicell, LA., Gnyp, A.W. and St. Pierre, C. 1986. A mathematical aaalysis of &or threshold detemninations. Transactions of the Bir and Waste Management Association, Pïttsburg, P e n x m y l . , USA, 167- 183.

Niceii, J. 1994. Development of the odor impact model as a regulatory strategy. Lnt. Journal of Envlronment and Pollution, 4: 124138.

Noren, 0. 1986. Design and use of biofflters for livestock buildings. In: Odor prevention and control of organlc sludge and llvestock £armlng. Elsier Applied Science Publishers, London, England, 284238.

O'Brien, M.A. 1093. Guidelines for odor sampïing and measurement by dilution ol£actometry. EE-6 subcommittee on the

standardization of odor measurement, Air and Waete Management Association. PIttsburg, PA.

O'Brien, MA. 1998. Revlsed guidelines for odor s a m p m and measurement by dynamic dilution olfbctornetrg. EE-6 mbcomrnittee on the stAnchrâization of odor meamrement, Air and Waste Management Association. Pittsburg, PA, USA.

O'Brien, MA., Duffee, RA. and Ostajic, N. 1996. Effect of sample flow rate in the determination of odour threshold. EE-6 mbcoxnmittee on the stsndardization of odor measurement, Air and Waste Management Association. Pittsburg, PA, USA.

O'Neill, D.H. and Phillips, V.R. 1992. A review of the control of odor nuisance from livestack buildings: Part 8, Properties of the odorous substances which have been identifled in lïvestock wastes or in the air around them. Journal of Agricultural Engineering Reaearch, 63: 23-50.

Patni, N.K. and Jui, P.Y. 1993. Effectiveness of m u r e additives. Paper no. 93402 1. A8AE, St Joseph Michigan, USA.

Ritter, W. F. 1989. Odor control of llvestock wastes : Sate-of-the-art in North America. Journal of AgriCUItural Engineering Reaearch 42:

51-62.

Schaefer , J. 197'7. Sampling characterization and analpis of malodors. Agric. and Env., 3: 121-12'1.

Schiffman, S. 1996. The effects of environmental odor emanating from commercial swine operatione on the mood of nearby residents. Brain Research Bulletin, 3'?(4): 369-376.

Schiffman, S. 1994. Physiological effects of swine odors on humans. in: Round Table Discussion on Odor Control, Iowa Univereity, Ames, Iowa, USA, 87-92.

Shurme, H.V. 1990. Basic limitations for an electronic nose. Sensors and Actuatore, BI, 48-68.

77

Sobel, AT., Ludington, D.C. and Yow, K.V. 1088. Altering âairy mnriure characteriatics for the aolld hanàling by the addition of becMing. Journal of 8011 and Water Conservation, 9: 132-137.

Sorel, J.E., Gauntt, R.O., Sweeten, J.H., Reddelï, D.L. and McFarland, A.R. 1983. Design of a 1-butanol scale àynamic ol£actometer for ambient odor measurements. Transactions of the ASAE, 26(O4) : 1201-1206.

Spoelstra, S.F. 1980. Origin of objectionable odorous components in piggery wastee and the possibiliQr of appying indicatior components fro studylng odor development. Agriculture and Environment, 5: 24 1-260.

Sweeten, J.M. 1996. Odor measurement technology and applications: a state of the art review. In: Seventh International Symposium on Agricultural and Food Proceseing Wastes. ASAE, St Joseph, Michigan, USA, 2 14-229.

Sweeten, J.M., ReddeU, D.L. and McFarland, A.R. 1983. Field measurement of ambient odors with a butesol olfactometer. Transactions of the ASAE, 26<04): 1206-1216.

Tandem Trade Corporation. 1996. Personaï conversation with Bk. Paul Jowàain. Evelyn Avenue, Toronto, Canada.

Thu, K-, DeLind, L., Dumenberger, E.P., Flora, C, Flora, d., Heffernan, W. and Paàgitt, S. 1996. Social issues. in: Under~tanàing the impact of large-male swine production. Proceedings from an Interdisciplhmqy Scientific Worlrshop. North Central Center for Rural Development, Tiie University of Iowa, Des Moines, Iowa, USA.

Watts, P.J., Jones, M. Lott, S.C., Tucker, S.W. and Smith, R.J. 1992. Odor measurement at a Queensland Feedlot. ASAE paper 024816. American Society of Agricultural Engineering, St Joseph, Michigan, USA.

78

Yaauhara, A.J. 1987. Identification of volatile compoundn in poultry m u T e by GC-M8. Journal of Chromatographie Science, 26: 37 1-878.

General Conclusion

The environmental impaat of the swine indwtry is closely related to the volume of fecee produced by pigs and the concentration of its undigested nutrients. S-er volumes of and les8 concentrated maniipes will reducea the problem related to land disposal, aver fertilization, and soi1 and water contamination. Aleo, potential odor production and g a ~ voiatiîization will be decreased. Thus, the moet logical wqy to 51d a sustainable solution ia by attacking the problem at its source, for instance, by improvlng the feed emciencg of the pig.

The present research showed that zeoïite (77% clinoptilolite) supplemented in grower hog rations improves feed efnciency. Zeolite ha8 been shown to be 84gdicant19 beneflcial when fed at a level of 2% to pigs weighing up to 40 kg, when compared to a aimlïar ration having the same energg and crude protein levels. Zeolite at level of 6% increases the feed efflciency and decreases the average daiïy consumption without meeting the rate of gain of pigs weighing more t h a n 50 kg. !ï'hus, by wing a zeoïite supplement at a level increaaing from 2 to 6% with the weight of the pigs, it is possible to reduce feed consumption by 13.6 kg per finished ho@. This represents a 6.8% reduction in feed consumption during the growing-finiah- phase. It al80 decreases the environmental impact of swine productions by decreasing the amount of manure produced. Abo, zeoUte tends to affect positively, but not significantly, the carcase q'IIAUty of elaughtered pigs for carcasses weighing les8 than 86 kg.

When cornParing the room where hogs were receiving the control feed containing sand and that where hogs were fed the zeoïite feed, no sipnificant differencee in ammonfa levels were found. However , dilut5.n.g the pig's feed wlth 2 or 6% of zeollte or sand reduced the armnonia 1-1 in the air of both rooms. Thus, it is possible to reduce the ammonia

level inside frrrm buildings by redudng the crude protein level of the feed. Very low levels of bydi.agen sulphide were measured throughout the experiment in both rooms. The level of carbon dioxide was not signiflcantly nifferent for both rooms. However, carbon dfoxide was directly Mected by the ventilation rate. Nonetheleas, paneïista found that the odor level of the zeollte room was slîghtly leas wîth the 2% zeolite than that of the control room, but there was no dieference with 5% zeolïte. The management and design of the ventilation system, the ventilation rate, the CleAnlInese of the room and feed crude protein had more sipnrflcant effects on air q W w and odor lm1 inside the grower rooms than the uee of zeoiite.

FinaUy, an automated dynamic olfactometer was conceived and built, based on the forced chaice Wanguhr method. Thia apparatus Fe innovative by its level of automation, it8 simplicity of operation, its rapidity of executfon and ite level of predsion. in f b t , the olactometer can calibrate six paneuste simulteneously with n-butanol and evaluate 4 dffferent odor samplee wi- 40 minutes. The user only needs to enter the required parameter on the setup wizard of the Windows oriented control software. AU other t a a u are autOmaticaUy performed by the olfactometer itself. The database oriented software Uows the user to print the results after the test or to keep them for M e r consultation. The b a t - i n butanol Lqlector removes the need of preparing butanol samples for quick and eaay callbration of panelist. This iqjector allows one to tramform the ol£actometer in a n-butanol male olîactometer, if need be. The oifbctometer is a good tool to measure odors in the agricultural sector,

5.1 Recomendations for Future Research The present research has demonstrated that zeollte ha& the

potential to reduce the environmenfal impact of the swine industry by increasing the admals ' feed efaciency. Neverthelesa, Airther research needs to be done in order to iden= it8 more specific effects, such as:

1. The optimal lwel of zeolite supplement in reference ta the

animal's weight. 2. Having found and using the ideal level of zeolite in temns of the animal's weight, a large male experiment ~hould be conducted ushg an 'aU in aïï outn m m . This system would provide more valuable and manger 8tatistical re8UIte.

3. A research inveetigating the effects of zeolite on the metabolisrne of pigs needs to be inveatigated. A nutrient mas8 'balance expertment needs to be perfomned uaing metabollsm cages to better understand how pigs utiïize zeolite- This research should determirie the effects on manuTe volume and nutrient content.

4. A atudg on the perception of agricuîtural odors should be done with the ol£actometer to flnd the detection tbreahold, the recognition threehold and tolerance tbeshold of variou8 odors. This study should provide the information required to derive a speciflc dilution mode1 for agrEuitura1 odor, taking into account the type of animais, feed, houeing and manure maaagement. This should lead to a etandard for the evaluation of a@icultural odors and give guidelines for a more apeciflc and reaiistic reguiation in thie field.

5. F-, it would be verg interesting to conceive and build a new generation of swine housing faciïîties with all the knowledge we poasesa in odor control and environmenîal impacts associated with the mine industrg. This ewfne housing facility should be used to verifSr the interactive effects on the various techniques used to reduce the soil, water and air pollution and enhance the well being of the pig, worker, nei@ibor and the entire commiuiity.

APPENDIB: A - 8tatistical Model

A. 1 Experimental Model The experlaaental model in this project was a temporal repeated-

measures ANOVA with a randomized complete block design. The experimental design was compoeed of four blocks with two to four replïcates per treatment per block, two treatmenta (control and zeoïîte), four (bi-weelrly) repeated measure and one covariable. The use of a covariable was necessarg in this experiment to remove Merences in the weights between groupa of pige. The covariable corresponds to the initial welghta of pi@ measured when the experiment atarted.

The blocks are there to increase the precision of the analysi~. They t e e into account the posaible spatial heterogenety due to the gradient in air qualit9 and the proxîmïw of the interior access (Figure A-1).

Figure A-1 Experimental layout; a) Treatment assignment, b) Block ass iment , c) Ventilation flow pattern

In order to anaïyze the data correctly, it was essential to extrapolate the missing values, which were calculated with 888 (Figure Cl). The data ha8 been verified to make sure it wrrs normellg distributed (Figure C-2). Moet of the results were normally distributed except for the time X, of AD1 in the 2% zeolite experiment. Therefore. no traneformatlon of the dsts -88 judged neceesary prior to a.nz4ysis.

The anaJysig of variances were performed on data concerning the average daily intahe, the average daQy gain and the feed/gain ratio to test the effects of zeollte for signifiamce. The classical A.NOVA, the modified ANOVA and the W O V A were three analyzes performed with SAS (Figure C-3) to test the nul1 hypotheses. The signif'icance levels were determined accoràîng ta Wilka' criterion and the F-values were considered eigniflcant when the probabiliw lm1 wes lese than 0.06.

A. 2 Classical ANOVA in this experiment, the cleasical ANOVA model -es variances

in the feedlgain ratio, average daily gain and amrage âaily intalce of pigs under two treatments. The ciassical ANOVA model is deacribed as:

with i = 1,2 ; j = 1,2,3,4 where p = population mean

4 = treatment main effects B, = block main effects (aB), = error term 1

E, = error term 2

The treatment fsctor is flxed and the block fbctor and the errors are random.

A.3 Modifîed ANOVA The modifled ANOVA model talces into account the correlation

between the dependent variables and considers tirne as a factor. The tfme is a withrn-subject factor end it ie crossed with treatment and with block.

The modified ANOVA model is described as :

with 1 = 1,2 ; j = 1,2,3,4 where p = population mean

b Y, =covariable main effects cr, = treatment main effects B, = block main effects (a), = error term 1 Ab, Y, = interaction of time t with covariable Y,

Ct = time main effects (aC),, = interaction of treatment f with time t (BC),, = interaction of treatment j with time t E, = emor term 2

The treatment and the tïme factors are flxed effects and the block factor and the errors m e random effects. The Greenhouse-GeWser (G-G) were used to W e the results in this experiment.

A.4 MANOVA The MANOVA model analyzes simultaneously the four dependent

miables (X,, &, and X4) called vectora of observations. The MANOVA model is deacribed as:

& t = m + P Y U + a + 8 , + & t

with i = 1.2 ; J = l,2,3,4 where g& = mU1tiVB]PI(Lte intercept

p Y, = multivariate cova,rïable main effecte a, = multivariste treatment main effects 8, = multivariate block main effects & = muhivariate error

The treatments are flxed effects and the block and the errors are random effecte. The Willf'a Lambda test was used for the amQmIS.

APPENDffL: B - Growth Performance Data

Table B-1 Average DaUy Gain <bW) et 2% zeoW

analyefe due to the excessive number of miesing date.

C C C C C C C C

C 1

!ïYeatment I Block ICovariabIe l x. I xm I r, I L 1 a a a a b b b c c

A

81.07 24.80 27.60 26.76 23.67 63.67 32.33 39-83 34.33

C Z Z Z

Note: the stricken column as been remove fkom the statisticai anelgels due to the exceseive number of miseing àete.

87

Table B-2 Average Daiîy Intake CADI> et 2% zeolite

30.33 27.00 22.17 24.67

d a a a

Treatment C C C C C C C C C C

0-G4 0.99 0.81 0.82 0.68 0.99 0.92 0.88 0.87

--IL401 >66-

/.- ,-

I -

_A -----

-/

-,û.SSt- >43%-- > t 5 Y _W

Z Z

Block a 8

a a b b b c c d

1 -- 1 -0

0.87. 1 0.62 0.67 1 0.66 0.41 0.61 O. 77 1 1.02 0.33 0.72 0.76 1 0.97 0.61 0.60 0.60 [ 0.67 0.62 1 0.76

--- --

--;-'

0.62 0.64 1 0-77

a f 26.00

t;ooarieBle 21.67 24-50 27.60 26.76 23.67 63.67 32.33 30.83 34.33 30.33

2 1 a

Z 2

0.69 1 0.83 0.62* 1 0.70 0.49 1 0.70 0.48 0.68 0.68 1 0.78 b

27.00 22.17 24.67 26 .O0 23.67 66.60 32.60 40.67 64.60 32.60

Z Z Z Z Z Z Z Z Z

Note: the stricken c o l m ha8 been removed h.om the statisicà,

0.66 I

0.76 O, 72 0.76 0.67

0.66 1 0.97. 0.66 I 0-69

I

0.60 I 0.86 0.49 1 0.76. 0.61 1 0.76

23.67

XI /-- ---.- _ _---

-,-.-- - --= _- -

-- -- --;-y

>.-7W - L W - , L . - 7 r -1.48- - - L l Y

a a a b b b c c d

0.96 0.73 0.94 0-69 0.61

b b

- -Al-'?/ _--- ------

S6.60 1 ,0243- 32.00 1 40.67 1 _-a- 34.60 -0.60- 32.SO 1 >Y

z Z Z

=4

1-98 2.41 2.16 2.48 1.87 3. 17

1.38 1.13'

x, 1.34' 1.60 1.28 1.7'1 1.21 2.42

c c d

xs 1.63 1.86 1-70 2.32 1.73 3.03 A

1-78 1.84 1-76 1.61

2.02 1.67

- 1 1.28 - i 0.98

1 1.32

2.18 1-99

1-80 1.80 1.79

3-11' 1.93 2.67 2.16.

- - L W /-.- /

- 3 - 7 C -,M--

2.04 2.06 2.08 3.76 2.17 3.04 2.26

1.86 2.17 2-18 1.88

2.09 1.63 1.80 1.46

1.83

2.22 2.66 2.74 2.08

>&W- 1.44 1 2 .O2

Table B-3 Feed to sain ratio (F/G) et 2% zeolite

due to the excessive number of rnissing data.

L

Z z Z Z Z Z Z

Treatment C C C C

Covariable 2167 24.60 27.60 28.76 23.6'1 63.67

Bloch a a a a

Note: the stricken column as been remove from the statistical m i 8

a b b b c c d

Table B-4 Average DaJly Gain (ADO) at 6% zeoïite

C C

26.00 83.67 66.60 32.6

40.67 34.60 32.60

Treatment C C

*1

/ / / / ,. /----

3.258-- b b

Bloch b b

CavaFiable 46.30 50.06

Z, 2.91 2.83 2.77 2.87 2.38 3-11

Z, 2.48. 2.22 3.10 2.22 3.68 S. 16

/

/---

,3.88/

/-

--2.C -J=€W-

4 8-64 2.41 2.63 3.03 2.73 3.17

XL

0.70 0.66

2.02 2-84 3.14 2.69 3.00 2.94 2.80

X, O. 76 0.82

2.64 2.ao 3-27' 3.23 2-97 2.82' 2.63

2.68 2-99

1

3.87 2.94 3.22 3.28 3.66

X, 0.76 0.63

x4

0.63 0.74

Table B-8 Average DaLly Intahe < D I ) et 6% zeolite

Treatment 2 . . . . . . . . . ZeoW gmups of pi@ C . . . . . . . . . Control groupe of pige

Block a, b, c, d . . . . Block, the epatial heterogenelty effect (Figure 15-1) Covariable Y . . . . . . . . . Repreeente the average initial weîght of a group of pi@ X1 -X, x . . . . . . . . . Repreeenta the two week interval where pîg8 are ~ 8 m e d

. . . . . . . . . . Ueingwalue extrapoletedby 8bS (Pygure C-1)

Treatment C

Table B-6 Feed to Gain ratio (F/G) at 5% zeolite Treatement

C

Block b

Blmk b

-18 46.30

&vari&le ¶

=1 x, x, 4 46.30 2.62 2-68 2-97 S.46

XL 1.86

4 2.20

X, 1.99

=S

2.24

APPENDIX C - 8AS Code

DsTa PIC); INPUT TRT $ BIDCK $ Y Xl-X4; CARDS;

ROC G I A NOPRINT; W S TRT MODEL 2U-X4=Y TRT B- OUTPUT OUT = PIGP

P = -1-XP4; PROC PRINT DA!I!A=PIQP;

i'igure C-1 888 code for the extrapohtion of the miaslng m u e

- - -

DA!I!A PI@ INPUT TRT 8 B m $ Y Xl-X4;

PROC GLM NOPRJNC; CLASS TRT BIDClQ MODEL Xl-X4=Y TRT B m OUTPUT OTJT =PIOR R=XRl-XE#; PROC 'ITNLVaRIPITE DA!i!A=PIGR NORMAL P m ; VAR aRl-aR4;

Figure C-2 SAB code ta test the normallQ~ of the data

DA!I!A PI* INPUT TRT $ B m $ Y X1-X4; CARDS ;

PROC GIX; CLASS TRT BLOCIE; MODEL X1-X4=Y TRT B m MANOVA H=TRT; MANOVA H=BfXK=g; REPEATED TIME 4 C O N T R B S T / ~ Y ;

C-3 BA8 code to test the hypotheses -th a mvariable

D a T a P I a INFUT TRT 8 BIAXE $ Y Xl-X4; CARDS;

$ROC GLM; CLAM TRT BUXIC; MODEL X1-X4= 'ITRT BUXIE; MANOVA H=TRT; w o v a H=BIXX=~; REPEATXD TiMZ 4 WNTRASWSUMMARY;

Figure C-4 8A8 oOde to te& the hypothesea without a cmmhble

- - -- -

DsTa Pm; INPUT TRT 8 BLOCg $ Y Xl-X4; X1 = X I E x 2 = m ; X4=X4/Y; CARDS;

PROC om; CLsSS TRT BLOCg; MODEL Xl-X4= TRT B m MANOVA H=TRT; W O V A H=BLOC%. REPEATED TiME 4 C O N T R A B T / ~ Y ;

iïgure C-5 SAS code to test fhe hypotheses by diviàing the data by a covarUb1e

Table D-2 S

** = ai--t et pc0.001 Mode1 A = statisticaï mode1 wl th covariable

B = wlthout c-ble, C = divldïng by covariable