To develop a categorical plumage scoring system for Zebra Finches, we divided the plumage surface into several focal parts to get a detailed picture and to identify potential problem areas of individuals. Therefore, we used the following six plumage areas: i) head, ii) neck, iii) back, iv) wings, v) tail, and vi) throat (Figure 1), in line with similar areas used in poultry. The area “head” is defined as the feathered part of the bird's skull which is adjacent to the “neck”. Neck is defined as the feathered part around the flexible part of the connection between body and head. Between the underneath of the beak and the ventral section of the “neck” is the region “throat”. From the dorsal “neck”, between the bird's “wings”, down to the preen gland is the area defined as “back”. This is followed by the “tail”. The wings of the bird and, in particular, the flight feathers are the part of the area “wing.” The so-far mentioned regions represent the obvious visible regions of the plumage which can, in theory, also be recorded without handling birds. However, it is recommended and implemented here to gently take birds into the hand to estimate the plumage scores. Other plumage regions turned out to be ill-suited, at least in the Zebra Finch, for different reasons, as they are either hard to assess in practice and some regions vary with sex and reproductive status and may therefore reveal no reliable well-being-related information. Finally, the sides and legs were not included.

At each of these six specified areas, plumage condition was categorized into one of four scores. The scoring categories were established in line with the procedures in laying hens (6). Thus, for each plumage area, one of four categorical values for that respective plumage status was assigned. Plumage categories were scored between I (worst) and IV (best). Category value IV is defined as a plumage in best condition, which means that the plumage is complete and dense, there are almost no significant gaps or damaged feathers (>95% undamaged, Figure 2), and flight feathers and tail feathers are complete and tidy. Category III means that the plumage is predominately complete with few gaps or damaged feathers. It can be fully feathered, but the feathers are tousled instead of smooth (>75% damaged). The flight feathers can show slight defects (category III, Figure 2). There are some of the smaller tail or wing feathers missing or the tails looks tousled. In category II, the number and dimension of gaps and damaged feathers increases (<75–>25% in undamaged status). For the body part “wing,” there are at least two flight feathers missing (category II, Figure 2), and for the body part “tail”, some of the long tail feathers are missing. Furthermore, a visible preen gland indicates a category II “tail”. The worst category is I. Here, the plumage contains big gaps, partly no feathers at all, highly damaged feathers, or even injured skin (<25% in undamaged status). At last, one wing shows a loss of one-third of its flight feathers (category I; Figure 2). A tail of the category I misses nearly all long tail feathers. By summing up the individual scores (head, neck, back, tail, wing, and throat), we get an overview of the bird's plumage condition. For each of the six body regions, exemplary pictures for each category are provided (Figure 2).

For the overall assessment of the plumage, we suggest a single value that allows a direct intuitive judgment of the feather condition, that is, the plumage damage score. To calculate the damage score of a bird, one subtracts the sum of the six bird's plumage scores from the maximum possible score (e.g., 24–21 = 3, the bird has a summed plumage score of 21 which means a damage score of 3). If required, each one of the six considered plumaged body regions can be analyzed separately. The lower the damage score, the better the bird's plumage. The above-mentioned scoring of the six different body regions can maximally result in a maximum of 24 score points (6 × category IV), or a minimum of 6 (6 × category I). Assuming feather scores of ≤ 2 at any single area indicates severe damage of the feathers, which means that the summed feather score of all six body regions cannot exceed 12 (6 × category II or lower); therefore, a damage score of 12 or higher is associated {[24–12 (or less) = 12 (or more)]} with severe feather damages. Similarly, an individual score > 3 and a total score of > 18–24 result in a damage score of 6 or lower [(24–18 (or more) = 6 (or less)] and would indicate very intact plumage.

In Zebra Finches, the subcutaneous fat stores are clearly visible and therefore quantifiable using appropriate scales. As Zebra Finches show a low amount of subcutaneous fat in general, we decided to adapt the score system used by Kaiser (23). We categorized the amount of visible fat on the ventral side of the birds (Figure 3), where especially the furcular and abdominal fat stores were of interest. In addition, we determined whether fat covers the wider abdominal and breast area. Birds with no visible fat in these focal areas received the score zero. Slight yellowish shades in the focal areas indicated a score of 1, clear yellow colouration indicated a score of 2 and slightly bulging fat deposited in one of the areas, or clear yellowish fat on both areas indicated a score of 3. Clear yellow colouration and a clear bulging of at least one area was scored as a 4. If the fat covered more areas of the abdomen and breast, this was scored as 5. The original score (23) included higher fat scores than 5. However, as the vast majority of Zebra Finches in our focal domesticated bird populations at Bielefeld University [referred to as “Bielefeld” in Forstmeier et al. (24); and “DOM Bielefeld” in Hoffman et al. (25)] are between 0 and a maximum of 4, we decided here against a further division above 5, which may, however, be implemented in other domesticated populations. Instead, as an addition to the originally suggested scoring scheme, we allowed a subdivision between two fat scores, that is, in 0.5 steps. This turned out to be very helpful, for example, if a bird shows more fat than for category 1, but not enough for category 2, thus we could assign a score of 1.5. This subdivision of the fat scores makes sense as one score reflects the entire body fat condition of the birds, while, for example, in the feather score, we assume that such a subdivision is not necessary, although easily possible, as there are already six different plumage scores, this is more subtle and reflects the bird's entire plumage condition.

Standardized and careful handling of the birds is important to generate reliable scores. The bird has to be held in one hand on its back. The neck is gently fixated between two fingers and the head is fixated carefully. The legs are held with the other hand. By gently puffing into the plumage, one can get a view of the bird's skin and the yellowish subcutaneous fat beneath.

To assess the reliability of our proposed scoring system, we conducted reliability measures at two levels, that is, the inter-observer reliability and the intra-observer reliability.

In total, 50 individual Zebra Finches were scored for the inter-observer reliability. About half of them were scored by 5 observers, and all of them by the two most experienced ones. To assess inter-observer reliability, 27 Zebra Finches (13 males, 14 females) were scored by five different observers independently within 10 min per bird. This short period was chosen to keep handling short for birds and to exclude differences in the bird's phenotype over longer periods such as days, weeks, or months. Observer #1 (L.K.) was most experienced, observer #2 was familiar with the scoring scheme, and the other tree observers #3–#5 were initially untrained with regard to the scoring scheme but received a short introduction of about 15 min from observer #1 in advance. However, all observers were very experienced with the general careful handling of Zebra Finches. The visual guide of the plumage scoring system (Figure 2) was available during data collection to all observers and also the illustrations of the fat scores. Additionally, the bird's body weight was taken after scoring to the nearest ± 0.01 g, using a digital balance (Kern EMB 600-2). Another 23 birds (2 females, 21 males), thus in sum N = 50 (16 females, 34 males), were scored by the observers #1 and #2 to assess the reliability with a larger sample size between the two most experienced observers.

For the reliability analysis, we did not calculate total plumage damage score, that is, the sum of all six plumage scores, but the six single scores of all six body parts were used, as this is a more conservative approach, which we assumed the best for the reliability analysis. This ensures that one can see exactly where the observers have scored different values, and subtle disagreements are not summed out (for further details refer to electronic Supplementary Table S1).

Statistical analyses were conducted using R (R version 4.1.0) (26). We analyzed the agreement of each possible observer pair with Spearman's rank correlation coefficients. Spearman's rank correlation coefficients r between 0 and 0.19 indicate a very week correlation, between 0.2 and 0.39 a weak correlation, between 0.4 and 0.69 a moderate correlation, between 0.7 and 0.89 a strong correlation, and between 0.9 and 1 a very strong correlation.

To test the intra-observer reliability, observer #1 (L.K.) scored 22 birds (10 males, 12 females) five times within 1 day, that is, 7 h. Between each scoring was at least a break of 60 min for each bird to calm down and to eat and drink. The time gap was also important to reduce potential memory from the previous scores from the observer's mind, which of course was not fully possible, but at each recording event, 132 measures (6 body regions in 22 birds) were taken, so that the chance of remembering single scores was quite moderate. Birds were assessed as randomly as possible between sessions. The intra-observer analysis was conducted similarly to the above-described inter-observer reliability using Spearman's rank correlation coefficients for all pairs of the five measurements. Graphs for the agreements were made using the R-package “hexbin” (27).

All birds were taken from the stock of and remained at the Department of Animal Behavior, Bielefeld University. The measures were taken to assess and to establish a reliable well-being assessment measure. All persons handling the birds were, in general, experienced with the gentle and careful handling of the birds. Score collection was carried out in accordance with the German laws and required no specific permit as the brief handling was estimated to not stress the animals equivalent to, or higher than, that caused by the introduction of a needle in the skin in accordance with good veterinary practice (Directive 2010/63/EU). Breeding and housing of the birds were done with the permission of the local authorities, the Veterinäramt Bielefeld (no. 530.421630-1, 18.4.2002, and no. 530.4, 27.07.2014). All birds always had ad libitum food and water available and were housed in groups of at least 4 birds in cages or small aviaries. All birds were monitored daily.

Our proposed plumage scores for Zebra Finches showed a substantial inter-observer reliability for the most experienced observer #1 in comparison with all four other observers (all r > 0.62, Table 1a). Considering reliability between all possible observer pairs, the Spearman's correlation coefficient was at least r > 0.56 (Table 1a). The average r-value of all comparisons was r = 0.642.

For the fat score, we obtained slightly different results. The Spearman's rank correlation coefficient showed lower values, which, however, were at least r > 0.39 when considering all possible observer pair comparisons (Table 1b). The average r-value of all comparisons was r = 0.514. Fat score seems to be a less reliable scoring system than the newly developed plumage feather score and it seems that it requires more experience to obtain a reliable measure.

The inter-observer reliability for the novel plumage score was substantial and it seems to be feasible to quick and easy teach inexperienced observers. In contrast, the fat score from Kaiser (23) seems to be a score for more experienced observers. Figure 4 illustrates the score agreements of selected observer pairs for plumage (Figures 4A–C) and fat (Figures 4D–F); all possible pairs are available in the electronic supplement.

The intra-observer agreement for the plumage scores was substantially strong (Table 2a). The Spearman's rank correlation for the measurement one vs. five showed a strong and significant agreement (r = 0.84, Table 2a). The average r-values of all comparisons of the five measurements were high with r = 0.852 (Table 2a).

For the fat score, a strong agreement (r = 0.90; Table 2b) was found when considering measurements one vs. five. The average r-value of all comparisons of the measurements was r = 0.90.

Figure 5 illustrates the score agreements of selected measurement pairs for plumage (Figures 5A–C) and fat (Figures 5D–F); all possible pairs of comparison are shown in the electronic supplement.