b i o s y s t em s e ng i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 5 3e2 6 0

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te / i ssn /15375110

Research Paper

Time derivatives in air temperature and enthalpyas non-invasive welfare indicators during longdistance animal transport

Morris Villarroel a,*, Pilar Barreiro b, Peter Kettlewell c, Marianne Farish c,Malcolm Mitchell c

aDepartment of Animal Science, School of Agricultural Engineering, Polytechnic University of Madrid,

Ciudad Universitaria s/n, 28040 Madrid, SpainbDepartment of Rural Engineering, School of Agricultural Engineering, Polytechnic University of Madrid, 28040 Madrid, SpaincSAC, The Roslin Institute Building, Easter Bush, Midlothian, EH25 9RG, UK

a r t i c l e i n f o

Article history:

Received 27 May 2010

Received in revised form

26 July 2011

Accepted 26 July 2011

Published online 10 September 2011

* Corresponding author. Tel.: þ34 914524869;E-mail addresses: morris.villarroel@upm

tlewell), Marianne.Farish@sac.ac.uk (M. Fari1537-5110/$ e see front matter ª 2011 IAgrEdoi:10.1016/j.biosystemseng.2011.07.011

Extreme environmental temperatures and high relative humidity can have serious nega-

tive effects on animal production at the farm level, but less is known about environmental

changes during live transport of domestic animals to slaughter. Although upper temper-

ature limits have been established to transport pigs in Europe, few indices include relative

or absolute humidity maxima or mention appropriate enthalpy ranges. In this study we

measured temperature, humidity and calculated air enthalpy (kg water kg dry air�1) on

commercial farms, during seven long distance (>24 h) journeys and at an abattoir. There

was an approximate overlap of data points on the psychrometric charts for each location

(farm, transport and abattoir). However, the temperature time derivative (�C s�1) and

enthalpy time derivative (kg water kg dry air�1 s�1) were up to ten times higher during

transport than the corresponding derivatives on the farm or at the abattoir. Post-transport

observation of pig behaviour also suggested that journeys with higher temperature or

enthalpy time derivatives were more stressed (evaluated as the amount of time they spent

resting or drinking). In conclusion, times derivatives of temperature or enthalpy could be

used as non-invasive welfare indicators during transport and appear to be much more

sensitive than absolute values of temperature or relative humidity.

ª 2011 IAgrE. Published by Elsevier Ltd. All rights reserved.

1. Introduction described about the effect of time derivatives of enthalpy on

Environmental temperature and relative humidity affect

animal production (Seedorf et al., 1998; Whittemore &

Kyriazakis, 2006; Zumbach et al., 2008), but little has been

fax: þ34 915491880..es (M. Villarroel), pilar.bsh), Malcolm.Mitchell@sa. Published by Elsevier Lt

pig welfare (Daskalov, Arvanitis, Pasgianos, & Sigrimis, 2006).

Presumably, at the farm level where animals spend most of

their lives, temporal changes in temperature and relative

humidity are moderate. However, during live transport,

arreiro@upm.es (P. Barreiro), Peter.Kettlewell@adas.co.uk (P. Ket-c.ac.uk (M. Mitchell).d. All rights reserved.

Nomenclature

Ldate loading date

Tavg average temperature, �CTmax average maximum temperature, �CTmin average minimum temperature, �CRHavg average relative humidity, %

RHmax average maximum relative humidity, %

RHmin average minimum relative humidity, %

Nsensor number of sensors on the livestock vehicle

Parea area of polygon

THI temperature humidity index

b i o s y s t em s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 5 3e2 6 0254

environmental conditions can change abruptly, which may

have more of an effect on animal welfare.

According to European Union Council Regulation 1/2005

(Chapter 2, Section 3.1, EuropeanCommission, 2004, pp. 1e44),

pigs in transport should not be subjected to temperatures

above 30 �C, and never above 35 �C. The thermal comfort zone

for 100 kg pigs is estimated to be 20 � 2 �C (Whittemore &

Kyriazakis, 2006). However, little attention has been given to

humidity values during transport, which may change more

drastically than temperature, having a direct effect on

enthalpy and the apparent temperature perceived by the

animals (Barbosa Filho et al., 2008).

Enthalpy is the heat energy of the air surrounding the

animal and dictates the degree of heat loss to the environ-

ment. According to the temperature humidity index (THI),

pigs are most comfortable at a THI lower than 75 (using �C,Lucas, Randall, & Meneses, 2000). However, relatively little is

known about time derivatives of enthalpy on the farm, and

how they may change during transport and at the abattoir.

Abrupt changes in the enthalpy pre-slaughtermay have direct

andmore serious effects onwelfare than degree differences in

temperature alone.

In this study our aim was to develop and compare

psychrometric charts from commercial farms, on livestock

transport vehicles and the destination abattoir using a large

data set from long transports (>8 h). Temperature time deriv-

atives (the change in temperature with time, �C s�1) and

enthalpy time derivative (H s�1) were also calculated and

related to behavioural data of pigs upon arrival, suggesting that

timederivativescanbeusedasnon-invasivewelfare indicators.

Table 1 e Summary of the seven journeys betweenScotland and Spain, including loading date (Ldate),average temperature throughout the journey (Tavg),average maximum temperature (Tmax), averageminimum temperature (Tmin), average relative humiditythroughout the journey (RHavg), average maximumrelative humidity (RHmax), average minimum relativehumidity (RHmin) and number of sensors on the livestockvehicle (Nsensor).

Trip Ldate Tmax Tavg Tmin RHmax RHavg RHmin Nsensor

1 04/06 28.11 18.23 10.20 90.43 58.90 23.65 4

2 09/07 29.80 20.37 13.78 91.20 66.00 29.68 4

3 23/07 40.64 22.97 16.19 84.50 58.79 23.90 4

4 06/08 35.29 21.84 15.04 90.50 63.84 23.55 4

5 20/08 32.86 20.16 12.55 92.10 65.49 23.60 4

6 18/09 31.83 16.90 7.43 86.65 59.58 28.53 4

7 15/10 24.52 15.11 5.02 94.82 65.62 36.67 10

2. Materials and methods

2.1. Journeys and experimental animals

Seven journeys were made from Scotland to Malaga, Spain

between the months of May and October in 2008 using

a commercial livestock transport vehicle carrying 80 pigs on

each trip. The loading and unloading dates and average

temperatures and humidity were all noted, as well as the

average temperatures inside and outside the vehicle during

transport (Table 1). Pigs were loaded in Edinburgh, trans-

shipped onto the instrumented truck at Ellenthorpe, York-

shire, UK and taken via a ferry crossing of the English Channel

from Poole to Fougeres, France, where they were unloaded

from the vehicle and rested for the mandatory 24 h period.

After that rest period, the pigs were reloaded and taken to an

abattoir in Humilladero, Malaga, Spain. All pigs were

Landrace� Largewhite, and approximately 100 kg liveweight.

The trailer measured 2.5 m by 8.0 m and had 3 levels, each

penmeasuring 2.4m by 2.4m. There was an average of 10 pigs

per pen (depending on live weight, sometimes 9 pigs), for an

average loading density of 180 kg m�2. Animals were kept in

stable groups at all times during housing, transport, rest

periods and lairage.

2.2. Sensors on farm, during transport and at abattoir

Data loggers (Hobo H8 loggers, Onset computers, MA, USA)

were used to measure temperature and relatively humidity

around the pigs before loading, during transport and at the

abattoir. Two sensors were placed on the farm in two pens

near the experimental animals and recorded once every

30 min from May to October. Two more sensors were placed

inside the lairage pens and also recorded once every 30 min

from May to October.

Prior to loading the vehicle, data loggers were mounted on

the partition gates between the pens on the middle deck (the

sensors were at pig height, 450 mm above the floor). The data

loggers were protected from direct contact and damage by the

pigs by housing them in a perforated steel framework and

were pre-programmed to record, at regular 2 min intervals.

Sensors on the truck were fitted and removed before and after

each journey. Approximately 20 mm of shavings were placed

on each floor of the vehicle as bedding.

The measurements of air temperature and relative

humidity were continuous throughout the whole transport

period from Ellenthorpe to Humilladero. When the pigs were

unloaded at Fougeres the data loggers remained on the

vehicle, however, the conditions experienced by the pigs were

b i o s y s t em s e ng i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 5 3e2 6 0 255

very similar to those on the empty truck because of the open

nature of the housing pens at the control post.

2.3. Data handling

The psychometric charts produced using the data collected

from the sensors were computed based on the ASABE model

which includes temperature, relative humidity, absolute

humidity and enthalpy. The psychrometric data ASAE D271.2,

defined in April 1979 and reviewed in 2005 (ASABE 2006 St.

Joseph, MI, USA) were used to calculate the psychrometric

properties of the air at farm, during transport (both outside

and inside the vehicle) and at the abattoir.

The temperature derivative was computed using the

Savitzky-Golay algorithm to smooth one-dimensional, tabu-

lated data and to help compute the numerical derivatives

using the Savgol routine in Matlab version 7.0 (Mathworks

Inc., Natick, MA, USA). Thus, a polynomial was used to fit the

data surrounding each data point. The smoothed points were

computed by replacing each data point with the value of its

fitted polynomial. Numerical derivatives arose after

computing the derivative of each fitted polynomial at each

data point. In our case a window of 21 points was used with

a five-order polynomial.

Next, the speed of change (with respect to a previous value)

or temperature derivative (dT/dt) and enthalpy derivative (dH/

dt) were calculated and plotted at each location. Finally,

calculating the temperature derivative produced different

polygons per sensor. Using Matlab, we calculated the area of

the polygons (Parea) that included all the data points in the

temperature derivative space per journey.

2.4. Behaviour analysis

Behaviours were assessed that might indicate how the pigs

coped with transport, changes in temperature and fatigue.

For each journey two trained observers monitored two

designated pens of 10 pigs (20 pigs behaviourally sampled per

trip) spray marked with a coding scheme on arrival. Pigs for

observation always came from the same pens on the truck

and were housed in the same pens in lairage. Pigs had access

to continuous water from one nipple drinker and an over-

head shower per pen. There was adequate space for all pigs

to lie laterally although no bedding was provided (lairage

pens measured approximately 5 m � 3 m). Pigs were not fed

in lairage. Behaviours were recorded continuously for 3 h

post-transport on each pen of 10 pigs simultaneously,

effectively resulting in 10 focal animals per pen. Close

attention was paid to posture changes and drinking behav-

iour. The behaviours recorded were mutually exclusive and

defined as; drinking: pigs actively intake water placing nipple

drinker in the mouth, licking or sucking water from the floor

or catching water openmouthed from overhead shower flow;

and resting: pigs were either in a still sitting position or

inactive lateral or ventral lying posture. Behavioural analysis

was drawn from these observations as; latency to drink (the

time the pigs waited after unloading to drink), frequency of

drinking (the amount of times each pig engaged in a drinking

bout), duration of drinking (total amount of time spent

drinking) and duration of resting (the total amount of time

sitting or lying).

Behaviour was recorded live using continuous sampling of

the pre-designated groups of pigs using the Psion ‘Work

About’ (Psion PLC, London, UK) hand held computer running

the Observer Package behavioural software version 3.0 (Nol-

dus, Wageningen, Netherlands). The data files from the ‘Work

Abouts’ were downloaded after each journey onto a laptop

computer using Noldus Observer 5 software.

3. Results

3.1. Journeys and experimental animals

All seven journeys were completed with no mortalities or

injuries to the pigs, between June and October 2008. The

average journey length, including the rest period at Fougeres,

was 63.3 h. In journeys 6 and 7, 77 and 79 pigs were loaded,

respectively, since some pigs were deemed unfit for transport

(pre-loading). Average temperatures were highest during

journeys 3 and 5, while average relative humidity was highest

during journey 4 (Table 1).

3.2. Temperature, humidity and enthalpy

Since temperature and humidity values were taken every

2min during transport, a total of 1900 data points (over 63.3 h)

were obtained for each journey. The changes in temperature

during transport are shown in Fig. 1 for journey 3, under some

of the hottest conditions.

According to the psychrometric charts obtained, the

enthalpy of the air surrounding the pigs at the farm in Scot-

land, during transport and at the abattoir in Spain largely

overlapped (Fig. 2). Although temperatures were slightly

lower in Edinburgh and higher in Malaga, the average

enthalpy values ranged between 0.005 and 0.02 kg water kg

dry air�1.

However, the temperature derivative was much higher

during transport than at the loading or unloading sites (Fig. 3).

The change in ambient temperature during transport varied

between �0.025 and 0.025 �C s�1 during transport, 10 times

higher than the range of changes at the farm (0.008 and

�0.001 �C s�1) or abattoir, (0.008 to �0.008 �C s�1).

Similarly, the enthalpy derivative (Fig. 4) was much higher

during transport, (0.08 to �0.08 kJ kg�1 [dry air] s�1), than at

loading (0.002 to �0.002 kJ kg�1 [dry air] s�1), or unloading

(0.0025 to �0.0015 kJ kg�1 [dry air] s�1).

As seen in Fig. 3, calculating the temperature derivative

produced different polygons per sensor, for which their areas

were calculated (Table 2). The area was smallest for trip 1, and

largest for trip 3, which corresponds with the range of

temperatures for those trips (also see Table 1), but also with

the speed of change in temperature.

With regards to current legislation, the number of minutes

that pigs were exposed to temperatures above 30 �C or 35 �Cwas calculated (Table 3). Temperatures were above 30 �C on

four journeys (3, 4, 5 and 6) but only above 35 �C for more than

1 min on trip 3.

Fig. 2 e Psychometric charts on (a) farm before loading in Edinburgh, Scotland, with the two different colours referring to

two separate sensors, (b) during transport inside the livestock vehicle from Scotland to Spain, where different colours are

different journeys, and (c) at the abattoir in Humilladero, Malaga, Spain, where the two colours are also two sensors. The

relative phase space of enthalpy values during transport roughly overlaps with those normally experienced by the pigs on

the farm. The different colours on the charts refer to different sensors at those locations.

Fig. 1 e Changes in temperature throughout long distance journey 3, during some of the hottest weather. Note the

important decrease in temperature when pigs were off-loaded at Fougeres (after 18-h transport), a staging post, and the

following increase when reloaded 12 h later.

b i o s y s t em s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 5 3e2 6 0256

Fig. 3 e Summary of the temperature time derivative on (a) farm before loading in Scotland (Edinburgh), (b) during transport

inside the livestock vehicle from Scotland to Spain, and (c) at the abattoir in Humilladero, Malaga, Spain. Note that the

ordinate values are 10 times higher during transport, indicating that the time derivative in temperature during transport is

10 times higher than on the farm.

b i o s y s t em s e ng i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 5 3e2 6 0 257

3.3. Behaviour analysis

After unloading, pigs from journey 3 spent more time drinking

than pigs in the other journeys (Fig. 5), while after most other

journeys the pigs spent more time resting. The duration of

drinkingwasalsomuchlonger for thepigs fromjourney3,while

latency to drink was similar among the pigs from all journeys.

4. Discussion

In this study temperature and humidity values at the farm,

during transport and at an abattoir were noted, and the

temperature and enthalpy derivatives were calculated, which

appeared to affect the behaviour of animals post-transport.

Our results suggest that the derivatives in temperature and

enthalpy during transport provide a much more sensitive

non-invasive indicator of animal welfare than temperature

alone. According to the psychometric charts, air enthalpy did

not vary widely at the loading or unloading sites or even

during transport, indicating that temperature or relative

humidity values by themselves are not very sensitive indica-

tors of environmental stress.

Abbott et al. (1995) reported an increase in pig mortality in

hot humid conditions and Mota-Rojas et al. (2006) concluded

that long distance transport during summer increases stress.

However, Gosalvez, Averos, Valdelvira, and Herranz (2006)

found an increase in mortality after transporting pigs in

autumn in Spain, compared to summer months, although

they provided no temperature or humidity data.

Fig. 4 e Summary of the enthalpy time derivative (with respect to its previous value) on (a) farm before loading in Edinburgh,

Scotland, (b) during transport inside the livestock vehicle from Scotland to Spain, and (c) at the abattoir in Humilladero,

Malaga, Spain.

Table 2 e Summary of the areas of the polygons (Parea)that included all the data points in the temperaturederivative space per journey. Each area is an average ofthe four sensors (N ) on the vehicle (except for trip 7whichhad 10 sensors). A low area implies a gradual change intemperature, even though the range in temperature mayhave been high. The %Max is area of that trip divided bythemaximumarea found among all sensors for all trips (asensor in trip 3). SE is the standard error for area (SEarea) ormaximum percentage (SE%max).

Trip N Parea (�C2 s�1) SEarea %Max SE%max

1 4 0.071 0.02 13.8 4.7

2 4 0.102 0.02 19.7 4.7

3 4 0.389 0.02 75.3 4.7

4 4 0.120 0.02 23.1 4.7

5 4 0.151 0.02 29.2 4.7

6 4 0.133 0.02 25.8 4.7

7 10 0.101 0.01 19.5 2.9

b i o s y s t em s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 5 3e2 6 0258

Although there are well-established guidelines regarding

enthalpy values and appropriate ranges in pig production

(Whittemore & Kyriazakis, 2006), less information is available

on changes in enthalpy with time (enthalpy derivative),

probably since derivatives are quite low in normal farm

environments. During transport, however, we found that both

Table 3 e Summary of the number of minutes that pigswere exposed to temperatures above 30 �C (>30 �C) andabove 35 �C (>35 �C), as well as the ratio between thevariance in temperature values inside and outside thelivestock vehicle.

Trip >30 �C >35 �C

1 0 0

2 0 0

3 203 50

4 338 1

5 297 0

6 66 0

7 0 0

0.05.0

10.015.020.025.030.035.040.045.0

1

2

3

45

6

7

Frequency drink Duration rest (mins)Duration drink (mins)Latency to drink (mins)

Fig. 5 e Behaviour of pigs after transport in terms of

frequency of drinking, mean duration of resting, mean

duration of drinking and mean latency to drink.

Table 4eAverage number of minutes that pigs were overthe limits suggested for two temperature humidityindices (THI, see text) used to assess possible effects onenvironmental temperature and humidity on pigwelfare.

Journey THI1 (NWSCR, 1976) THI2 (Ingram, 1965)

�84 �79 >¼75 �84 �79 >¼75

Emergency Danger Alert Emergency Danger Alert

1 0 0 0 0 0 0

2 0 0 0 0 0 0

3 0 6 46 34 52 104

4 0 0 4 2 18 298

5 0 0 0 0 0 188

6 0 0 0 0 0 32

7 0 0 0 0 0 0

b i o s y s t em s e ng i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 5 3e2 6 0 259

temperature and enthalpy derivatives varied much more

rapidly. Our exhaustive analysis of temperature and relative

humidity on seven journeys (13,300 data points in total),

suggests that both temperature and enthalpy values increase

or decrease 10 times faster during transport compared to the

farm or abattoir, approximately 0.6 �Cmin�1 during transport,

compared to 0.06 �Cmin�1 on the farm or abattoir. In addition,

when more relative change was experienced by the pigs, they

spent more time drinking after transport, indicating that

those conditions were more stressful. Other environmental

stressors such as noise are also typically higher during

transport than at the farm (Talling, Lines, Wathes, & Waran,

1998), underlining the high amount of variation found in

livestock transport vehicles.

Our data onmaximum temperatures (Table 3), agreed with

the principles behind EU Council Regulation 1/2005 (European

Commission, 2004), since the pigs that spent more time above

30 �C or 35 �C also appeared to be more stressed. However, it

should be underlined that enthalpy values were a more

accurate predictor of pig stress than the total number of

minutes over 30 �C, which does not provide an idea of varia-

tion or humidity.

Lucas, Randall, and Meneses (2000) compared two temper-

ature humidity indices (THIs), one based on NWSCR (1976)

where I1 ¼ 0.72tw þ 0.72td þ 40.6, and another based on Ingram

(1965) where I2 ¼ 0.63tw þ 1.17td þ 32. When the results of the

seven journeysdescribed in thispaper (Table 4)were compared,

many more pigs were in an emergency situation based on I2than with I1. Based on our results and personal observations,

THI1 appears tobeaccuratemeasurement ofpig stress since it is

more sensitive and better reflects problem journeys.

One of the most important practical implications of this

study is that the same temperature sensors currently being

usedonanimal livestock vehicles canbeusedasmore sensitive

sensors by calculating the temperature derivative. Although EU

Council Regulation 1/2005 (European Commission, 2004) only

mentions a general range of maximum and minimum

temperatures, our data point out the usefulness of calculating

temperature derivatives than can be easily obtained using the

same source data and are linked to welfare indicators. Adding

humidity sensors to moving vehicles is often complex since

sensors may get wet and may have to be removed before

disinfection after unloading. In addition, those sensors are

usuallymore costly than temperature sensors alone. According

to our data, calculating the temperature derivative may be

sensitive enough that humidity sensors are not required,

thereby reducing costs.

Acknowledgements

We wish to thank Eddie Harper for technical assistance and

Dr. Tim King and staff at the LAU, Roslin Institute UK, as well

as the helpful staff at the Matadero de Humilladero. This study

was funded by the UK Government Defra (Department for

Environment, Food and Rural Affairs), project number

AW0820, “Transcontinental road transport of breeder pigs e

effects of hot climates”.

r e f e r e n c e s

Abbott, T. A., Guise, H. J., Hunter, E. J., Penny, R. H. C., Baynes, P. J., &Easby, C. (1995). Factors influencing pig deaths during transit:an analysis of drivers’ reports. Animal Welfare, 4, 29e40.

Barbosa Filho, J. A. D., Correia Vieira, F. M., Fonseca, B. H., OliveiraSilva, I. J., Brito, D., & Hildebrand A. (2008). Poultry transportmicroclimate analysis through enthalpy comfort index (ECI):a seasonal assessment. In: Livestock Environment VIII, 31Auguste4 September 2008, Iguassu Falls, Brazil. 701P0408.

Daskalov, P. I., Arvanitis, K. G., Pasgianos, G. D., & Sigrimis, N. A.(2006). Non-linear adaptive temperature and humidity controlin animal buildings. Biosystems Engineering, 93, 1e24.

European Commission. (2004). Council Regulation (EC) No 1/2005 of22 December 2004 on the protection of animals during transport andrelated operations and amending Directives 64/432/EEC and 93/119/EC and Regulation (EC) No 1255/97. OJ L 3, 5.1.2005, pp. 1e44.

Gosalvez, L. F., Averos, X., Valdelvira, J. J., & Herranz, A. (2006).Influence of season, distance and mixed loads on the physicaland carcass integrity of pigs transported to slaughter. MeatScience, 73, 553e558.

Ingram, D. L. (1965). Evaporative cooling in the pig. Nature, 207,415e416.

Lucas, E. M., Randall, J. M., & Meneses, J. F. (2000). Potential forevaporative cooling during heat stress periods in pig

b i o s y s t em s e n g i n e e r i n g 1 1 0 ( 2 0 1 1 ) 2 5 3e2 6 0260

production in Portugal (Alentejo). Journal of AgriculturalEngineering Research, 76, 363e371.

Mota-Rojas, D., Becerril, M., Lemus, C., Sanchez, P., Gonzalez, M.,Olmos Ramırez, S. A., et al. (2006). Effects of mid-summertransport duration on pre- and post-slaughter performanceand pork quality in Mexico. Meat Science, 73, 404e412.

NWSCR. (1976). Livestock hot weather stress: regional operationsmanual letter C-31-76. USA: National Weather Service CentralRegion.

Seedorf, J., Hartung, J., Schroder, M., Linkert, K. H., Pedersen, S.,Takai, H., et al. (1998). Temperature and moisture conditions

in livestock buildings in Northern Europe. Journal ofAgricultural Engineering Research, 70, 49e57.

Talling, J. C., Lines, J. A., Wathes, C. M., & Waran, N. K. (1998). Theacoustic environment of the domestic pig. Journal ofAgricultural Engineering Research, 71, 1e12.

Whittemore, C. T., & Kyriazakis, I. (2006). Whittemore’s science andpractise of pig production. Oxford (UK): Blackwell Publishing Ltd.

Zumbach, B., Misztal, I., Tsuruta, S., Sanchez, J. P., Azain, M.,Herring, W., et al. (2008). Genetic components of heat stress infinishing pigs: development of a heat load function. Journal ofAnimal Science, 86, 2082e2088.