The use of portable X-ray fluorescence spectrometry to measure apparent total tract digestibility in beef cattle and sheep

Abstract

The use of portable X-ray fluorescence (PXRF) spectrometry to detect external markers on processed or unprocessed cattle and sheep fecal specimens to estimate apparent total tract digestibility (ATTD) was evaluated. Exp. 1: ruminally cannulated Angus-crossbred steers (n = 7; BW = 520 ± 30 kg) were individually fed ad libitum for 21 d in a completely randomized design (CRD). Markers (Cr₂O₃ and TiO₂) were placed inside the rumen twice daily (7.5 g of each marker). Fecal samples were collected twice daily from day 14 to 21. Exp. 2: crossbred wethers (n = 8; BW = 68 ± 3 kg) were individually fed ad libitum for 21 d in a CRD. During this period, 2 g of Cr₂O₃ and TiO₂ were top-dressed onto the feed twice daily. Sheep were housed in metabolism crates for 5 d for total fecal collection. Concentration of markers was determined on diets, refusals, and fecal specimens (fresh, dry-only, and dried/ground) using atomic absorption to detect Cr and spectrophotometry for Ti. Concentration of both markers was also determined via the PXRF spectrometer. Delta between ATTD estimated by wet chemistry and PXRF was not different from zero (P ≥ 0.14) when using cattle fresh fecal specimens for both markers, whereas ATTD estimated by PXRF with dry-only and dried/ground fecal specimens were 3.6 and 1.1 percent units lower (P ≤ 0.04), respectively, than ATTD estimated by wet chemistry for Cr and Ti, respectively. Regardless of the fecal sample preparation method on cattle specimens, Ti concentration was similar (P = 0.39) among methodologies, while Cr was underestimated (P < 0.01) by 13% when PXRF was used in dry-only or dried/ground samples. The ATTD of sheep was underestimated (P < 0.01) by 2.4 percent units compared with control when Cr was measured by PXRF in dry-only samples. The Cr concentration in dry-only fecal specimens of sheep tended (P = 0.09) to be lower compared with wet chemistry analysis. Fresh and dry/ground sheep fecal samples assessed for Cr, and dry-only assessed for Ti were not (P ≥ 0.49) affected by detection method. The Cr fecal recovery tended (P = 0.10) to be the lowest for dry-only, the greatest for wet chemistry, intermediate for fresh and dry/ground sheep-fecal specimens; while not affected (P = 0.40) for Ti. The PXRF is an accurate technology to detect Cr and Ti in fresh cattle fecal samples to estimate ATTD. For fresh and dry/ground, the technology was effective for determining the concentration of Cr, or dry-only fecal specimens when detecting Ti in sheep specimens.

Introduction

The use of novel technological tools within the livestock industry is constantly increasing. For example, radio-frequency identification and global positioning system technology are used to monitor grazing patterns and behavior to consistently feed animals and monitor health status (Turner et al., 2000). This technology is also used in commercial feedlots to monitor the time in which pens are consistently fed and guarantee the proper diet goes to the correct pen. In addition, automatic machines can dilute, emulsify, and dispense micro-ingredients with greater accuracy and precision inside the mixer. All such technologies can increase the precision in nutrient management, but their efficacy will rely on the ability of the animals to digest and absorb the ingested feed. However, the beef industry is currently lacking an approach that provides a real-time measurement of nutrient utilization. A common way to predict the utilization of nutrients by the animals is currently performed by predictions of intake and measurements of body weight (BW) (NASEM, 2016). Although the prediction or actual measurements of feed intake provides great information for feeding management, it does not provide any direct information regarding the extent of digestion of the consumed nutrients. Moreover, measurements of BW are not performed routinely or frequently enough, and as a result, opportunities to detect subpar performance is less than ideal and may even lead to the detection of potential issues only after economic losses have already occurred. Therefore, the assessment of digestibility may grant commercial operations the ability to measure overall dietary nutrient utilization after dietary or management changes, and potentially detect issues before animal growth performance is affected. The first established method to estimate apparent total tract digestibility was through the use of total fecal collections (Lindahl, 1963); however, this method can be time consuming and inconvenient for research, and impractical for commercial operations. The use of dietary external and internal markers (Faichney, 1975), which can be determined by wet chemistry in the laboratory, brought to light a newer concept, in which total fecal collections could be replaced by spot fecal samples. Although less laborious compared with total fecal collections, the use of markers still requires time-consuming wet chemistry procedures to determine the concentration of said markers. The use of portable X-ray fluorescence (PXRF) spectrometry has been increasingly adopted within environmental and soil sciences for determining concentrations of various elements in soils and compost (Weindorf et al., 2008, 2012a, 2012b,; Zhu et al., 2011). Interestingly, two commonly used dietary external markers, titanium dioxide (TiO₂; Titgemeyer et al., 2001) and chromic oxide (Cr₂O₃; Barnicoat, 1945; Lloyd et al., 1955), contain elements that can be detected by PXRF spectrometry. Barnett et al. (2016) used PXRF to determine concentrations of Cr and Ti in cattle fecal samples that were dried and subsequently ground. We hypothesized that this technology is accurate enough to be applied in research and commercial settings for the determination of apparent total tract digestibility (ATTD). However, the process of drying and grinding samples is not a routine practice in commercial operations. Therefore, the current experimental objectives were to determine the level of accuracy that could be achieved using a PXRF device to measure concentrations of two common external dietary markers (Cr and Ti) in sheep and beef cattle fecal samples for three levels of sample preparation (as-collected [fresh], dried, or dried and ground) and the subsequent accuracy for prediction of ATTD.

Materials and Methods

All experimental procedures were conducted in accordance with approved Texas Tech University Institutional Animal Care and Use Committee forms (IACUC nos.: 19018-02 and 19026-03).

Experiment 1: design, treatments, and cattle management

Seven ruminally cannulated Angus-crossbred steers (BW = 520 ± 30 kg) were used in a completely randomized design. Steers were housed, individually, in the Ruminant Nutrition Center located at the Texas Tech University—New Deal Research and Education Center in Idalou, TX. This indoor facility is composed of individual concrete stalls (2.89 × 5.58 m), drainage system, automatic water troughs, heating/cooling system, and a chute for handling animals during collections if needed. The facility was washed with pressurized water twice daily. Steers were fed a steam-flaked corn-based diet (Table 1) at approximately 1.05 times ad libitum consumption (dry matter [DM] basis) once daily (0700 hours) for the entire duration of experiment. Diets were mixed every 4 d and stored in lid-sealed plastic containers under refrigeration (4 °C). Each marker was pre-weighed (7.5 g, as-is basis) into biodegradable gel capsules (Torpac, size 07, Fairfield, NJ). The external markers used were chromic oxide (Cr₂O₃) and titanium dioxide (TiO₂). The markers were introduced into the rumen through the ruminal cannula twice daily (at feeding and 10 h after feeding) throughout the entire duration of the experiment, totaling 15 g of each marker per d. Cattle were dosed with external markers for 14 d to reach a steady state, with subsequent collections performed for 7 d while continuing marker dosing. Both markers were individually utilized to estimate apparent total tract digestibility. Two analytical methodologies were compared: 1) traditional wet chemistry and 2) PXRF spectrometry. Wet chemistry analyses of fecal samples included atomic absorption (Shimadzu AA-6300, Columbia, MD 21046) for Cr (AA), and spectrophotometry (BioTek Epoch, Winook, Vermont) for Ti (Spec.), which were compared with PXRF spectrometry (Olympus Delta Premium-6000, Waltham, MA). These analyses were used to determine the concentration of external markers used on three types of fecal specimens analyzed by the PXRF device: 1) fresh (AVG DM = 22.5% ± 2.2); 2) dry-only, dried at 60 °C for 72 h; and 3) dried/ground, dried at 60 °C for 72 h followed by grinding to 1 mm (Wiley mill, Thomas Scientific, Swedesboro, NJ).

Table 1.

Dietary ingredients and analyzed nutrient composition of finishing diet fed to cattle (experiment 1) and sheep (experiment 2)

Item	Cattle	Sheep
Inclusion, % DM basis
Corn grain, steam-flaked	64.90	38.80
Wet corn gluten feed (Sweet Bran)1	20.00	38.53
Alfalfa hay, low quality	8.00	15.37
Yellow grease	3.00	2.50
Limestone	1.60	1.50
Urea	0.50	—
Ammonium carbonate	—	1.00
Supplement	2.002	2.303
Analyzed nutritional composition, % DM
DM4, % as-fed	76.88	73.82
OM4	94.08	93.25
CP4	14.72	15.12
NDF4	22.62	21.28
ADF4	8.79	7.72
EE5	4.00	4.61
Ca5	0.70	0.69
P5	0.48	0.50
K5	0.97	1.11
Na5	0.19	0.21

¹Cargill Corn Milling, Blair, NE.

²Supplement contained (DM basis): 67.7538% carrier (cottonseed meal), 0.5% antioxidant (Endox; Kemin Industries, Inc., Des Moines, IA), 3.76% urea, 10% potassium chloride, 15% sodium chloride, 0.0022% cobalt carbonate, 0.1965% copper sulfate, 0.0833% iron sulfate, 0.0031% ethylenediamine dihydroiodide, 0.167% manganous oxide, 0.125% selenium premix (0.2% Se), 0.9859% zinc sulfate, 0.0099% vitamin A (1,000,000 IU/g), and 0.157% vitamin E (500 IU/g) and provided (dietary) 30 mg/kg of monensin (0.75% Rumensin-90 in supplement; Elanco Animal Health, Indianapolis, IN) and 9 mg/kg of tylosin (0.5063% Tylan-40 in supplement; Elanco Animal Health).

³Supplement contained (DM basis): 71.69% carrier (corn), 0.5% antioxidant (Endox; Kemin Industries, Inc., Des Moines, IA) 2.59% dicalcium phosphate, 8% potassium chloride, 3.558% magnesium oxide, 12% salt, 0.00173% cobalt carbonate, 0.1333% iron sulfate, 0.002515% ethylenediamine dihydroiodide, 0.26667% manganous oxide, 0.1% selenium premix (0.2% Se), 0.84507% zinc sulfate, 0.5% thiamine hydrochloride, 0.00792% vitamin A (1,000,000 IU/g), and 0.125999% vitamin E (500 IU/g).

⁴Analyzed composition in house at the Ruminant Nutrition Laboratory.

⁵Analyzed composition from commercial laboratory (Servi-Tech Laboratories, Amarillo, TX).

Apparent total tract digestibility

During collection days (days 15 to 21) samples of diet were collected daily (approximately 200 g, as-is basis), and fresh fecal samples (approximately 100 g, as-is basis) were collected twice daily (0700 and 1700 hours) directly from the rectum upon spontaneous defecation or via rectal palpation. Individual time-point fecal samples were frozen at −20 °C for further analysis. Individual fecal samples were thawed over-night, composites were prepared within animal and day, and fresh daily-composited fecal samples were first scanned with PXRF device. The PXRF scanning followed the procedures of USEPA method 6200 (USEPA, 2007). The PXRF device utilized was equipped with a Rh X-ray tube, and the power setting was 10– to 40 kV in Geochem mode. Prior to scanning, the instrument was calibrated with a 316 standard alloy coin. For the unprocessed (fresh) fecal sample scanning, daily-composited fecal samples were placed in an aluminum foil pan (7.5 × 15 cm), and the PXRF lens was positioned in direct contact with fecal sample for scanning (90 s). The aperture of the PXRF device was fitted with a prolene thin film to protect the instrument from moisture and other foreign contaminants from samples. Upon completion of PXRF scanning, fecal samples in aluminum pans (approximately 300 g, as-is) were dried in a forced-air oven for 72 h at 60 °C. After drying, samples were removed from the oven and carefully separated from the aluminum pans (revealing a brick shape of dried feces; 7 × 15 × 3 cm). Samples (brick shape) were then divided in half to expose the internal portion to be scanned with the PXRF spectrometer, which followed identical steps performed on the fresh fecal samples. Finally, after the second scanning, fecal samples were ground to pass a 1 mm screen, homogenized, and submitted to the third scanning with the PXRF spectrometer. All wet chemistry determination of markers in diet and fecal samples was performed using dried and ground samples (the same samples used for the third PXRF scanning procedure). Concentration of Cr was determined by combination of AOAC (1997) Method 985.35. Briefly, pre-ground (1 mm; fecal and diet) sample composites were weighed in duplicate (5 g, as-is) into acid-washed porcelain crucibles. Samples were then dried in a forced-air oven at 100 °C for 16 h and subsequently ashed in a muffle furnace at 550 °C for 24 h: the final ash residue was weighed and recorded. On a hot plate, deionized water and concentrated nitric acid (20% HNO₃) were added in a 1:1 ratio in the crucible with the ashed sample, continuously, until all particles were digested. The residue sample was filtered (Whatman grade 40), rinsed with deionized water, and transferred to 50-mL volumetric flasks, then brought to volume with deionized water. Samples and standards (0, 50, 100, 150, 200 µL Cr standard solution Sigma-Aldrich, St. Louis, MO) were vortexed and absorbance of Cr read on an atomic absorption (AA) spectrophotometer (Shimadzu AA-6300, Columbia, MD 21046). Titanium dioxide was determined by the method described by Titgemeyer (1997). Briefly, pre-ground (1 mm; fecal and diet) sample composites were weighed in duplicate (0.5 g, as-is) into acid-washed porcelain crucibles and ashed for 3 h at 550 °C. The resultant ash residue and crucible were placed on a hot plate, and 10 mL of 7.4 M sulfuric acid was added, followed by a minimum 30 min digestion (or until sample was clear, not milky colored). Contents were allowed to cool and diluted with deionized water into 120-mL plastic cups, containing 10 mL of 30% hydrogen peroxide. The sample cups were then brought to 100 g weight with deionized water and filtered (Whatman No. 41) into 50 mL plastic tubes. Samples and standards (0, 1, 2, 4, 6, 8, and 10 mg TiO₂/100 g) were read with a microplate spectrophotometer (BioTek Epoch, Winook, Vermont) at 405 nm. Apparent total tract digestibility of DM was calculated by: [(1 − (Cr or Ti concentration in DMI / Cr or Ti concentration in feces) × 100] (Adapted from Church, 1975). The daily intake of Cr or Ti was calculated as marker supplied daily into the rumen plus the trace amounts of naturally occurring marker determined from the feed consumed daily.

Experiment 2: design, treatments, and sheep management

Eight crossbred wether sheep (BW = 68 ± 3 kg) were used in a completely randomized design. Sheep were housed, individually, throughout the entire duration of the experiment at the Texas Tech University—Animal and Food Sciences Livestock Arena in pens (length = 152.4 cm and width = 91.44 cm) and metabolism crates (length = 91.5 cm, width = 53.4 cm, and height = 78 cm). The experiment lasted 21 d, with an initial 14 d of adaptation to diets and markers where sheep were kept in individual dirt-floor pens. Following adaptation, sheep were moved to metabolism crates (length = 91.5 cm, width = 53.4 cm, and height = 78 cm) for 7 d, in which 2 d were used solely as an adaptation to crates, and the remaining 5 d were used for collections. Sheep were fed a steam-flaked corn-based diet (Table 1) at approximately 1.02 times ad libitum consumption divided in two daily meals (0700 and 1700 hours), except during the collection period, in which 95% of the ad libitum amount observed during the initial 14 d was fed daily. Diets were mixed every 5 d and stored in lid-sealed plastic containers under refrigeration (4 °C). Each metabolism crate was equipped with an individual waterer, feeder, and each animal was fitted with a fecal collection harness (canvas). Waterers were cleaned and refilled with fresh water twice daily (0700 and 1700 hours). The same analytical methodologies previously described in experiment 1 were also used in experiment 2 (traditional wet chemistry and PXRF), with the addition of a third comparison (control = total fecal collection). The three types of fecal specimens were also assessed: fresh, average 39.6 ± 5.5; % DM, dry-only, 60 °C for 72 h, or dried/ground, 60 °C for 72 h/1 mm. For fresh sheep fecal samples, the granular constituency of sheep fecal specimens was disrupted by thorough mixing to facilitate homogenization.

Total fecal collection and apparent total tract digestibility

Fecal collection harnesses were emptied, feces weighed, and fecal content collected (65% of total, fresh basis) twice daily (0800 and 1700 hours). As in experiment 1, both Cr₂O₃ and TiO₂ were used in experiment 2 to estimate ATTD. Each marker was pre-weighed (2 g, as-is basis) into sealed plastic bags, totaling 4 g (as-is basis) of each marker per day. Markers were mixed with wet corn gluten feed (5 g as-is, WCGF—quality Sweet Bran [Cargill Corn Milling, Blair, NE]) and top-dressed onto diet immediately prior to feedings. During collections (days 17 to 21) samples of diet and refusals (only trace amounts present) were collected daily, and fecal samples were collected twice daily (0800 and 1700 hours) from fecal collection harnesses. Samples were frozen at −20°C for further analysis. Diets, refusals, and feces were composited within day and animal. Thawed daily homogenized (to disrupt original spherical format and eliminate air pockets) composite samples (first PXRF scan, fresh) were dried at 60 °C for 72 h in a forced-air oven (second PXRF scan, dry-only). After drying, samples were ground to pass a 1-mm screen (third PXRF scan, dried/ground) in a Wiley mill (Thomas Scientific, Swedesboro, NJ). Measured total fecal output (DM basis) was used to calculate ATTD and compared with ATTD calculation obtained from concentrations of the external marker as previously described in experiment 1.

Statistical analyses

Data were analyzed using two complementary approaches: ANOVA and correlation/regression. For the first statistical approach, ANOVA was carried out using the GLIMMIX procedures of SAS (SAS Inst. Inc., Cary, NC). For experiment 1, within each type of dietary external marker (Cr₂O₃ or TiO₂), the fixed effect of treatment (three types of fecal sample [fresh, dry-only, and dried/ground] analyzed by the PXRF, plus the wet chemistry method for each respective marker), collection day, and interaction (treatment × day) were evaluated for all variables measured. For experiment 2, within each type of dietary external marker (Cr₂O₃ or TiO₂), the fixed effect of treatment (three types of fecal sample [fresh, dry-only, and dried/ground] analyzed by the PXRF, wet chemistry method for each respective marker, and total fecal collection method), collection day, and interaction (treatment × day) were evaluated for all variables measured. For both experiments, animal was considered the experimental unit; animal within treatment was included as a random effect; and collection day was considered the repeated measure. Covariance structure for repeated measures were chosen based on the smallest Akaike’s information criterion. The delta variables (the PXRF value obtained minus the wet chemistry value obtained [experiment 1] or minus the total fecal collection value obtained [experiment 2]) were deemed different from zero if P ≤ 0.05 using pair-wise comparisons. When the interaction term was not significant (P > 0.05), it was removed from the model. The Kenward–Rogers general degrees of freedom procedure was used to adjust for any bias on standard errors caused by multiple terms in the random statement.

The second statistical analyses performed used JMP (JMP Pro 14, SAS Inst. Inc., Cary, NC) by correlation and regression. The correlation analysis performed was among marker concentrations (Cr₂O₃ and TiO₂) detected in cattle and sheep feces, in which the three variations of PXRF fecal scanning (fresh, dry-only, and dried/ground) were individually compared against common laboratory methods (AA for Cr₂O₃ and Spec. for TiO₂). Additionally, the use of bivariate and regression platforms was performed to determine the correlation amongst ATTD estimates from the three variations of PXRF fecal scanning (fresh, dry-only, and dried/ground) and were individually compared against laboratory methods for experiment 1 and total fecal collection for experiment 2. The coefficient of determination (r²) was determined to be strong if values were 0.7 to 1.0, moderate if 0.4 to 0.7, and weak if less than 0.4. Significant differences were declared if P ≤ 0.05 and tendencies declared if 0.05 > P ≤ 0.10.

Results

Experiment 1: cattle

When Cr₂O₃ was used as external marker, no treatment × day interaction was observed (P = 0.59) for ATTD; it was consistent through days of assessment (P = 0.17) and not different (P = 0.47) amongst fecal specimen types analyzed by PXRF or wet chemistry (Table 2), averaging 79.8%. However, a more sensitive analysis calculating the ATTD delta (ATTD value obtained from PXRF scanning minus ATTD obtained from wet chemistry) when Cr₂O₃ was used as a marker revealed a treatment × day interaction (P = 0.02; Figure 1). For all days of digestibility measurement (except for days 5 and 6), ATTD was underestimated (P < 0.01) by 3.6 and 3.7 percent units when PXRF spectrometer was used to scan dry-only and dried/ground fecal specimens, respectively, compared with the wet chemistry, whereas fresh fecal samples scanned by PXRF did not differ (P = 0.90) in ATTD compared with the wet chemistry (Table 2). No treatment × day interaction (P = 0.11) or day effect (P = 0.27) was observed for concentration of Cr in the feces, whereas Cr measured by PXRF scanning of fresh fecal samples resulted in similar (P = 0.90) Cr concentration compared with the wet chemistry, while dry-only and dried/ground samples underestimated (P < 0.01) Cr fecal concentration by 0.09 and 0.13 percentage units.

Table 2.

ATTD and fecal external marker concentration measured on processed and unprocessed cattle fecal specimens analyzed by PXRF spectrometry or traditional wet chemistry

	PXRF						P-values
	Fresh	Dry	dry/ ground	AA1	Spec.2	SEM3	trt	d 4	trt × d
ATTD, % DM
Cr2O3	81.2	78.2	78.1	81.7	—	2.23	0.47	0.17	0.59
TiO2	83.4	83.1	83.0	—	84.1	1.64	0.89	0.3	0.04
ATTD delta5, % DM
Cr2O3	−0.5a	−3.6b	−3.7b	0a	—	0.87	<0.01	<0.01	0.02
TiO2	−0.8	−1.1	−1.2	—	0	0.70	0.36	0.01	<0.01
Marker fecal concentration, % DM
Cr	0.77a	0.69b	0.65b	0.77a	—	0.02	<0.01	0.27	0.11
Ti	0.77	0.77	0.73	—	0.77	0.03	0.38	<0.16	0.09
Marker fecal concentration delta5, % DM
Cr	−0.01a	−0.09b	−0.13b	0a	—		<0.01	<0.01	0.29
Ti	0.01	0.01	−0.04	—	0	0.022	0.36	< 0.01	0.17

¹Atomic Absorption (Shimadzu AA-6300, Columbia, MD) using dry (60 °C; 72 h) and ground (1 mm) fecal specimens.

²Spectrophotometer (BioTek Epoch, Winook, Vermont) using dry (60 °C; 72 h) and ground (1 mm) fecal specimens.

³Standard error of the mean (n = 7 per treatment “trt”).

⁴Days collected used for repeated measures.

⁵Delta refers to the difference between the digestibility coefficient observed when using PXRF minus the values obtained from the traditional wet chemistry (AA or Spec.).

Figure 1.

Experiment 1: cattle. Delta for ATTD of dry matter calculated from chromium (Cr₂O₃) detected by PXRF device or wet chemistry (AA) in beef steers fecal specimens. A treatment × day interaction was observed (P = 0.02), in which all days of digestibility measurement, ATTD was underestimated (P < 0.01) by 3.6 and 3.7 percent units when PXRF device was used to scan dry-only and dried/ground fecal specimens, respectively, compared with the wet chemistry (AA), whereas fresh fecal samples scanned by PXRF did not (P = 0.90) underestimate ATTD compared with the wet chemistry (except on days 5 and 6).

Except for days 5 and 6 of the collection period (treatment × day interaction, P ≤ 0.04, Figure 2; and day, P = 0.01, Table 2), when TiO₂ was used as an external marker in beef cattle finishing diets, the ATTD averaged 83.4%. The ATTD did not differ among treatments (P ≥ 0.36).

Figure 2.

Experiment 1: cattle. Delta for ATTD of dry matter calculated from titanium (TiO₂) detected by PXRF device or wet chemistry (Spec.) on beef steers fecal specimens. A treatment × day interaction (P ≤ 0.04) was observed, in which PXRF measurement on dry-only samples was different (P ≤ 0.05) than wet chemistry (Spec.) on days 2 and 3, while for days 5 and 6, PXRF analysis from fresh and dry/ground fecal samples were lower (P ≤ 0.05) than wet chemistry calculated ATTD.

When Cr₂O₃ was used as an external marker, it showed high positive correlations between the PXRF fresh and dried/ground fecal samples (r² = 0.83 and 0.76, respectively; Table 3) compared with Cr concentrations derived from AA. However, when AA was compared against PXRF dried fecal specimens, a moderate correlation was observed (r² = 0.65), which similarly corresponds with the ATTD moderate correlation (r² = 0.67). Compared with common laboratory methods with TiO₂ as a marker, PXRF detection of Ti in fresh feces showed high correlation (r² = 0.84), and significantly weaker correlations for PXRF dried and dried/ground feces (r²= 0.69, 0.50), respectively.

Table 3.

Correlation of marker concentrations and estimated ATTD between laboratory analyses and variously processed PXRF spectrometry cattle fecal specimens

Item	r 2	Summary of fit equation
Fecal marker concentrations
Cr2O3
PXRF_Fresh	0.8284	PXRF_Fresh (counts per second) = 3160.7341 − 0.1865283* fecal marker concentration (measured by AA1)
PXRF_Dry	0.6457	PXRF_Dry (counts per second) = 841.56798 + 0.1338916* fecal marker concentration (measured by AA)
PXRF_DryGr	0.7591	PXRF_DryGr (counts per second) = 1473.9564 + 0.6842848* fecal marker concentration (measured by AA)
TiO2
PXRF_Fresh	0.8363	PXRF_Fresh (counts per second) = 2657.184 − 0.1400902*fecal marker concentration (measured by Spec.2)
PXRF_Dry	0.6926	PXRF_Dry (counts per second) = 11406.974 − 0.6135355*fecal marker concentration (measured by Spec.)
PXRF_DryGr	0.4975	PXRF_DryGr (counts per second) = 7379.9538 − 0.1236634*fecal marker concentration (measured by Spec.)
ATTD
Cr2O3
Dig_FreshPXRF	0.6059	Dig_FreshPXRF = 8.758157 + 0.893166*Dig_AA
Dig_DryPXRF	0.6686	Dig_DryPXRF = -9.986358 + 1.1064853*Dig_ AA
Dig_DryGrPXRF	0.7052	Dig_DryGrPXRF = −2.716493 + 0.9970897*Dig_ AA
TiO2
Dig_FreshPXRF	0.7611	Dig_FreshPXRF = 14.55866 + 0.8276674*Dig_Spec
Dig_DryPXRF	0.4877	Dig_DryPXRF = 0.6965579 + 0.9928415*Dig_Spec
Dig_DryGrPXRF	0.5192	Dig_DryGrPXRF = 24.64832 + 0.6963157*Dig_Spec

¹Atomic Absorption (Shimadzu AA-6300, Columbia, MD) using dry (60 °C; 72 h) and ground (1 mm) fecal specimens.

²Spectrophotometer (BioTek Epoch, Winook, Vermont) using dry (60 °C; 72 h) and ground (1 mm) fecal specimens.

Experiment 2: sheep

The use of PXRF in sheep fecal specimens did not interact with day of collection (treatment × day; P ≥ 0.37) for any of the variables assessed herein (Table 4). There was a day effect (P ≤ 0.02), which justifies the use of multiple collection days, and as it did not interact with treatment, treatment × day interactions will not be discussed further for experiment 2. When Cr₂O₃ was used as external marker analyzed by the PXRF spectrometer, the level of fecal specimen processing seemed to influence the accuracy of the determination of ATTD. When PXRF was used to estimate ATTD using dry-only fecal specimens, a decreased (P < 0.01) digestion coefficient was observed compared with the other treatments, with a tendency (P = 0.09; data not shown) to also be reduced compared with the control method (total fecal collection). The delta assessment of ATTD (Table 4) was also affected by treatment where dry-only fecal specimens scanned by the PXRF tended (P = 0.09; data not shown) to underestimate ATTD by 2.36 percent units compared with control, which was less intense compared with the underestimation (P = 0.04) induced by the AA method (2.82 percent units compared with control). However, fresh and dried/ground fecal specimens scanned by PXRF did not differ (P ≥ 0.29; data not shown) from the control method, with a maximum numerical difference of 1.4 percent units in ATTD (Table 4). Fecal Cr concentration tended (P = 0.09) to differ among treatments; however, the fecal Cr delta (P = 0.03) showed that PXRF scanning of fresh sheep fecal samples yielded similar values compared with AA, whereas an underestimation (0.32 percent units) was observed for dried-only fecal samples. The Cr fecal recovery tended (P = 0.10) to be the lowest (86.7%) when dry-only fecal samples were scanned with PXRF, and to be the greatest (115.1%), when traditional wet chemistry was used, while PXRF scanning fresh fecal samples (101.9%) and dried/ground (94.5%) samples tended (P = 0.10) to yield intermediate values that were closer to 100% recovery compared with the value obtained by the other treatments (Table 4).

Table 4.

ATTD, fecal external marker concentration, and fecal external marker recovery rate measured on processed or unprocessed sheep fecal specimens analyzed by PXRF spectrometry or traditional wet chemistry

		PXRF						P-values
	Con1	Fresh	Dry	Dry/ ground	AA2	Spec.3	SEM4	trt	d 5	trt × d
ATTD, % DM
Cr2O3	75.67ab	77.12a	72.01b	75.08ab	78.88a	—	1.006	<0.01	0.02	0.93
TiO2	75.25	79.45	76.10	77.60	—	75.65	2.481	0.75	<0.01	0.82
ATTD delta6, % DM
Cr2O3	0bc	1.40ab	−2.36c	−0.03bc	2.82a	—	0.927	<0.01	0.50	0.55
TiO2	0	4.20	0.85	2.35	—	0.4	1.847	0.49	<0.01	0.60
Marker fecal concentration, % DM
Cr	—	1.02	0.86	0.94	1.18	—	0.162	0.09	<0.01	0.57
Ti	—	1.27	1.09	1.17	—	1.07	0.149	0.76	<0.01	0.86
Marker fecal concentration delta7, % DM
Cr	—	−0.16ab	−0.32b	−0.24ab	0.00a	—	0.072	0.03	0.25	0.79
Ti	—	0.21a	0.02b	0.09ab	—	0.00b	0.039	<0.01	0.02	0.37
Marker recovery rate, %
Cr	—	101.9	86.7	94.5	115.1	—	8.02	0.10	0.79	0.64
Ti	—	130.7	111.8	122.2	—	112.4	8.95	0.40	<0.01	0.43

¹Control: ATTD measured by the gold standard total fecal collection method.

²Atomic Absorption (Shimadzu AA-6300, Columbia, MD) using dry (60 °C; 72 h) and ground (1 mm) fecal specimens.

³Spectrophotometer (BioTek Epoch, Winook, Vermont) using dry (60 °C; 72 h) and ground (1 mm) fecal specimens.

⁴Standard error of the mean (n = 8 per treatment “trt”).

⁵Days collected used for repeated measures.

⁶Delta here refers to the difference between the digestibility coefficients observed when using PXRF minus the values obtained from the gold standard total fecal collection method.

⁷Delta here refers to the difference between the digestibility coefficient observed when using PXRF minus the values obtained from the traditional wet chemistry (AA or Spec.).

When TiO₂ was used as an external marker in sheep finishing diets, the ATTD averaged 76.81% (DM basis), which did not differ among treatments (P ≥ 0.75), ranging from 75.25% (control) to 79.45% (PXRF scanning fresh fecal specimens). The delta in ATTD was not affected by treatment (P = 0.49), although numerical overestimation of ATTD ranged from 4.20 to 0.4 percent units, for PXRF scanning fresh specimens and values obtained by wet chemistry, respectively. Nonetheless, the fecal Ti delta showed that PXRF scanning of fresh sheep fecal samples underestimated (P < 0.01) Ti concentration by 0.21 percentage units compared with wet chemistry, whereas dry-only samples yielded the closest value, with dried/ground samples being intermediate (Table 4). The Ti fecal recovery was not affected (P = 0.40) by treatment, ranging from 112% to 131% (Table 4).

When Cr₂O₃ was used as an external marker, it showed high positive correlations between all PXRF fecal scanning (r² = 0.84, 0.76, and 0.83 for fresh, dried, and dried/ground, respectively; Table 5) compared with Cr concentrations derived from AA. Additionally, the ATTD correlations compared with total fecal collections showed high positive correlations, regardless of PXRF fecal specimen processing (r² = 0.74, 0.71, and 0.73 for fresh, dried, and dried/ground, respectively; Table 5). Compared with common laboratory methods with TiO₂ as a marker, PXRF detection of Ti in fresh, dried, and dried/ground feces showed high correlations (r² = 0.84, 0.82, and 0.83, respectively). Apparent total tract digestibility estimated by PXRF scanning of Ti on fresh and dried/ground feces showed high correlation (r² = 0.71 and 0.70, respectively), whereas PXRF performed on dried feces showed moderate correlation (r² = 0.61) compared with total fecal output (Table 5).

Table 5.

Correlation of marker concentrations and estimated ATTD between laboratory analysis, total fecal collections, and variously processed PXRF spectrometry sheep fecal specimens

Item	r 2	Summary of fit equation
Fecal marker concentrations
Cr2O3
PXRF_Fresh	0.8360	PXRF_Fresh (counts per second) = 377.1057 + 0.2954606*fecal marker concentration (measured by AA1)
PXRF_Dry	0.7577	PXRF_Dry (counts per second) = 2071.2271 + 0.5308071*fecal marker concentration (measured by AA)
PXRF_DryGr	0.8260	PXRF_DryGr (counts per second) = 2457.9968 + 0.5734683*fecal marker concentration (measured by AA)
TiO2
PXRF_Fresh	0.8426	PXRF_Fresh (counts per second) = -1627.358 + 0.6324618*fecal marker concentration (measured by Spec.2)
PXRF_Dry	0.8189	PXRF_Dry (counts per second) = 1248.5334 + 0.8227833*fecal marker concentration (measured by Spec.)
PXRF_DryGr	0.8280	PXRF_DryGr (counts per second) = 1707.3466 + 0.8544902*fecal marker concentration (measured by Spec.)
ATTD
Cr2O3
Dig_FreshPXRF	0.7403	Dig_FreshPXRF = -9.656709 + 1.1409841*Dig_Total
Dig_DryPXRF	0.7053	Dig_DryPXRF = −16.2172 + 1.1876991*Dig_Total
Dig_DryGrPXRF	0.7336	Dig_DryGrPXRF = −8.074443 + 1.1060095*Dig_Total
TiO2
Dig_FreshPXRF	0.7110	Dig_FreshPXRF = −s1.990862 + 1.1003664*Dig_Total
Dig_DryPXRF	0.6095	Dig_DryPXRF = 10.849956 + 0.9024169*Dig_Total
Dig_DryGrPXRF	0.6979	Dig_DryGrPXRF = 10.692592 + 0.9256278*Dig_Total

¹Atomic Absorption (Shimadzu AA-6300, Columbia, MD) using dry (60 °C; 72 h) and ground (1 mm) fecal specimens.

²Spectrophotometer (BioTek Epoch, Winook, Vermont) using dry (60 °C; 72 h) and ground (1 mm) fecal specimens.

Discussion

Apparent total tract digestibility associated with animal intake is considered valuable information that can yield accurate predictions of animal growth and performance. Conventional methods for digestibility experiments, such as the ones conducted in the current study, have been used by researchers to compare the extent of digestion affected by feeding strategies, diets, or ingredients. The ATTD estimated utilizing these methods, specifically in beef steers consuming a high grain finishing diet, has been reported by numerous researchers (Zinn, 1992; Theurer et al., 1999; Montgomery et al., 2004). The ATTD ranged from 74% to 84 % in beef cattle consuming finishing-type diets, with steam-flaked corn as the primary concentrate source, concurring with our study.

In regard to the methodology, the concentration of Cr and Ti detected by PXRF in unprocessed and processed cattle fecal specimens was similar to laboratory-derived values. This is in agreement with observations made by Barnett et al. (2016) who reported positive correlations (r²= 0.93) in the detection of Ti with PXRF spectrometry compared with an inductively coupled plasma (ICP) method when spot samples of feces (dried and ground at 1 mm) were collected from cattle. It was observed in the current study that fresh fecal specimens from cattle might yield less variable measurements when using the PXRF technique and that these measurements may allow for more accurate estimations of ATTD, as shown by r² ≥ 0.83. However, the success in using fresh fecal specimens to detect indigestible markers appears to be directly related to the physical properties of feces, specifically the moisture content. Several parameters are known to affect PXRF accuracy, among them are particle size, surface irregularity, and moisture. Moisture is the most influential source of error for the device, especially when above 20% (Ge et al., 1997). These limitations were reported when PXRF was used in soils research. In addition, Weindorf et al. 2014b reported that the formation of water biofilms or layers on scanned samples by PXRF is more likely to induce loss of accuracy in the reading than the moisture content itself. Fecal particle size and physical consistency induced by the absence of regular shapes or surfaces in the fecal matter may have contributed to some of the variation in the current data. As seen within the current study, regardless of animal species and marker, it is reasonable to hypothesize that these factors could explain the less accurate readings of fresh sheep fecal samples scanned with the PXRF spectrometer, due to the different consistency of the feces compared with cattle, even though both sheep and cattle specimens contained more than 60% moisture. Nevertheless, the high precision noted for the PXRF spectrometer for fresh cattle fecal samples could be potentially attributed to several factors. For instance, the fecal sample having a greater moisture content (22.5% DM for cattle vs. 39.6% DM for sheep) permitted easier homogeneous mixing of the fresh sample. As reviewed and discussed by Peinado et al. (2010) and Weindorf et al. (2014b), the equal distribution, and mixing, of soil and compost samples can inherently affect the PXRF spectrometer sensitivity to detect the elements. In cattle, the capability of the PXRF to estimate ATTD appears to be effective for fresh fecal samples, when compared with laboratory methods.

In experiment 2, the accuracy to detect Ti increased for PXRF readings of dry-only fecal samples, while for Cr, the unprocessed (fresh) and processed (dried/ground) were more suitable. In accordance with the current data, Barnett et al. (2016) also reported success when detecting Cr in dried and ground fecal content of sheep (r² = 0.95) when comparing the novel and classical technique of ICP. Similar results were observed in the current study, in which Cr was successfully detected by PXRF (r² ≥ 0.76 and 0.82 for Cr and Ti, respectively) when compared with common laboratory techniques, whereas, comparatively, the detection of marker using PXRF in sheep seemed to respond differently than that of cattle in regard to similar fecal sample type (dry). The PXRF analysis on dried/ground and fresh sheep feces offered less variability in the prediction of ATTD than dry in relation to laboratory methods, which agrees with the findings of Barnett et al. (2016). For both experiments, ATTD determined by PXRF differed in physical fecal sample type and type of indigestible external marker used. In addition, even though the fecal pellets collected from sheep were disrupted during the mixing of the sample, there could have been air pockets that may have interfered with the PXRF readings. These air pockets could be the same causation for the low correlations (r² ≥ 0.52) associated with PXRF dried and grounded bovine feces used to estimate ATTD. As it has been investigated within soil and compost, sample physical form and moisture can affect the accuracy of the PXRF technique (Ge et al., 1997; Peinado et al., 2010; Weindorf et al., 2014a, 2014b). In the current study, it appears that the moisture within feces, regardless of species and marker utilized, allowed for more even distribution of particles to be detected with PXRF (r² ≥ 0.83). Additionally, the difference in the detection of markers (Cr and Ti) using PXRF between species in the current study could be thought to be attributed to the difference in marker distribution due to marker solubility. Differences identified on delta measurements on days 5 and 6 for the cattle experiment can be directly related to additional measurements also taken on those days (additional fecal specimens that were not used in current assessment).

Implications

The ability of the PXRF spectrometer to detect Cr and Ti in fecal samples, and consequently to be used to estimate ATTD, seems to be affected by physical characteristics of fecal samples, which are specific to animal species. Simply processing fecal specimens by dehydration and grinding to a specific granulometry does not seem to be an effective sample preparation method for all types of fecal specimens. For beef cattle consuming finishing diets, fresh fecal samples scanned with PXRF device yielded acceptable estimations of ATTD and fecal marker concentrations, when either Cr₂O₃ or TiO₂ was used. For sheep consuming finishing diets, when using Cr₂O₃ as an external marker, it seemed that PXRF scanning can be performed on either fresh or dry/ground samples, whereas, when using TiO₂, dried-only fecal specimens would be preferable. The PXRF spectrometry method seems to be a viable alternative to the traditional fresh chemistry techniques to estimate ATTD, providing faster and less expensive analysis than traditional techniques. Furthermore, it has the potential to be used on-site with results available immediately upon scanning. However, further investigation is warranted to identify specific factors that can influence the effectiveness of the device for use in specimens from other animal species, other dietary markers, and other sampling techniques (individual vs. commingling pens).

Glossary

Abbreviations

AA

atomic absorption spectrophotometer method for laboratory analysis of Cr

ATTD

apparent total tract digestibility

BW

body weight

Cr₂O₃

chromic oxide external marker, also abbreviated as Cr

CRD

completely randomized design

ICP

inductively coupled plasma

PXRF

portable X-ray fluorescence spectrometry

Spec

spectrophotometry method for laboratory analysis of Ti

TiO₂

titanium dioxide external marker, also abbreviated as Ti

Conflict of interest statement

The authors declare no real or perceived conflicts of interest.