The animals used in this experiment were bred at the Department of Behavioral Biology at the University Münster from a breeding program of multi-colored shorthaired guinea pigs (Cavia aperea f. porcellus) that were originally acquired from a breeder in 1975. All animals were born in small groups of one male, two to three females, and their pre-weaned offspring.

The used animals were weaned at about 21 days of age and transferred to either same sex groups (females; 2–12 individuals) or same-sex pairs (males) until the start of the experiment. During the experiment, each male was housed together with two females. They were housed in enclosures measuring 1 m × 1 m, with walls constructed of wood with a plastic red section at the bottom, in total 0.5 m high. The floor was covered with wood shavings (Tierwohl Super), and food (Höveler Meerschweinchenfutter 10700) and water with ascorbic acid was available ad-libitum. Hay was replenished daily. Each enclosure contained two shelters; one large red plastic shelter and one small red plastic or black wooden shelter. Lights were on daily from 7:00 to 19:00 and room conditions were ~22°C with about 50% humidity. At the end of the experiment, females remained at the Department of Behavioral Biology. Therefore, it could be verified whether or not the females became pregnant during the experiment. All females of groups with intact and sham-operated males became pregnant. All females housed together with castrated males did not become pregnant; females of 10 out of 14 groups with sterilized males, however, became pregnant (one sterilized male mated first 1 year later with females). Probably, a recanalization occurred in these animals. This is also known from other studies; however, to a much lower extent [e.g., (51, 52)]. One factor for a successful sterilization is the surgical procedure. Sterilization failure is observed more often when sterilization is performed without cauterization. In our study, only ligation, but no cauterization was used (see below, Section 2.4). For this reason, we had to exclude these 10 males from our analysis, since we could not know at what time recanalization occurred. Thus, we could analyze data of only 4 sterilized males. In future, it is strongly recommended to perform sterilization with cauterization and ligation.

The males had been castrated (n = 14), sterilized (14, only 4 remained sterilized) or sham-operated (n = 8) 30–45 days after birth, i.e., several weeks before sexual maturity, which is reached in male guinea pigs at about 70 days of age. Furthermore, intact males (n = 10) were also included. The castration procedure involved castration through the scrotum with immediate ligation of the blood vessels and suturing of the scrotum after the testis were removed. The sterilization procedure that was used involved excising an 8–10 mm piece of each of the two vasa deferentia after performing ligations on both ends to keep the vas closed (for details see below Section 2.4). During the experiment each one male (age about 85 days of age; range: 78–92 days of age) was introduced into a new enclosure together with two females (age of females about 75 days of age; range: 55–107 days of age; age differences between two females of one group: 0–40 days). Since each two females were housed together with one male, the sample size of females was the following: females living together with a castrated male: n = 28, with a sham-operated male: n = 16, with a sterilized male: n = 8, with an intact male: n = 20.

At about 34 days of age (range: 27–41 days of age) a cortisol response test (CRT; see below) was performed on each male prior to surgery to ensure that there were no differences between the males that were later treated differently at this time. With the age of about 85 ± 7 days (around sexual maturity) each one male was introduced together with two females into a new enclosure. 5 ± 1 day before this introduction a further CRT was conducted to assess initial values. Two further CRTs were performed 8 ± 2 days and 21 days ± 2 after introduction into the groups to assess the ability to cope with this new situation in the long term. Furthermore, a blood sample of each male was taken just before as well as 2 h after introduction.

Blood samples were also taken from the females (see below for procedure): 1 day before the females were put to the male to determine baseline values, 3 h as well as 1 day after being put together with the male to determine the acute cortisol response, and about 1 week later and on the very last day to determine the long-term cortisol response.

Surgeries were conducted under mixed anesthesia with ketamine and medetomidine i.m. (Ketamin^®, Ceva Tiergesundheit GmbH, Düsseldorf, Germany, 60 mg/kg body weight; Domitor^®, Pfizer, Berlin, Germany, 0.25 mg/kg body weight). In addition, local anesthesia was applied using a 0.5% lidocaine solution (Xylocain^® 1%, AstraZeneca, Wedel, Germany, in physiological saline, 0.05–0.1 ml s.c. in each side of the scrotum). Effects of medetomidine were reversed using atipamezole (Antisedan^®, Pfizer, Berlin, Germany, 0.75–2 mg/kg body weight s.c.) after completion of surgery (at least 45 min after induction of anesthesia). Postoperative analgesia was provided by carprofen (Rimadyl^®, Pfizer, Berlin, Germany, 4 mg/kg body weight s.c., after induction of anesthesia).

Castration: The scrotum was opened on both sides with a small incision, the testicles and epididymis were placed in front of the scrotum, ligated with absorbable suture material and removed. The scrotum was sutured in two layers (peritoneum with muscle layers and skin).

Sterilization: The abdominal cavity was opened on both sides in the groin with a small incision through the skin and abdominal wall, the vas deferens were placed in front of the abdominal cavity, and a pair of forceps was slid underneath to stretch the vas deferens open. Two surgical knots were placed between the legs of the forceps. In addition, the area between the nodes was cut out (~5–10 mm). The surgical access was then also sutured in two layers (peritoneum with muscle and skin).

Sham operations: The scrotum or abdominal cavity was opened at the same location and then closed again as described.

In male and female guinea pigs blood samples were taken to determine cortisol concentrations. In males, furthermore testosterone concentrations were determined.

In short, guinea pigs were undisturbed for 1 h prior to the onset of the sampling procedure which always started at 9.00 a.m. The guinea pig was then removed from his/her enclosure, and a marginal ear vein was punctured to collect about 50 μl of blood within 3 min (cortisol) and further 100 μl within 6 min (testosterone). Guinea pigs show little defense behavior during the collection procedure, and no elevation of plasma cortisol occurs for about 5 min (36). Accordingly, all samples for determination of cortisol levels were collected within 3 min of entering the room to make sure that no elevation of cortisol had yet occurred. After that the guinea pig was weighed.

With male guinea pigs several cortisol reaction tests were conducted. During this test 3 blood samples were taken. The first sample was taken as described above at 9.00 a.m. Instead of returning the animal to its home enclosure, it was placed singly in a 1 m² novel enclosure without shelter, which is a stressor for guinea pigs and triggers an increase in circulating cortisol (42) until exactly 1 h after the first blood sample (at 10.00 a.m.) was collected. This process was subsequently repeated until three blood samples were taken over 2 h at 11.00 a.m. (baseline, after 1 h, after 2 h) and then the guinea pig was returned to the home enclosure. Cortisol concentrations were determined from each blood sample; testosterone concentrations were also determined from all samples collected before start of the test.

The blood plasma was extracted directly after sample collection via centrifugation and then frozen at −20°C. Once all samples were collected at the end of the experiment, the concentrations of cortisol and testosterone were determined in duplicate using enzyme-linked immunosorbent assays (Cortisol ELISA, RE52061, IBL International GmbH, Hamburg, Germany; intra- and inter-assay CVs: 3.0 and 3.5%; antibody cross-reactivity: cortisol 100%, prednisolone 30%, 11-desoxycortisol 7%, cortisone 4.2%, prednisone 2.5%, and corticosterone 1.4%; testosterone ELISA, RE52151, IBL International; intra- and inter-assay CVs: 4.6 and 5.7%; antibody cross-reactivity: testosterone 100%, 11β-OH-testosterone 8.7%, 11α-OH-testosterone 3.2%, and dihydrotestosterone 1.9%). The technical assistant who performed the assays was blinded to the experimental design. In addition, all samples from one male were analyzed in the same assay. Furthermore, it was ensured that samples from at least three differently treated males were analyzed in the same assay.

All statistics were calculated using SPSS for Windows (SPSS, Chicago, IL, USA version 28). R 4.0.0 (53) and RStudio (Posit, Boston, MA, USA, Version 2022.07.1+554) was used to create graphs. We conducted an a priori sample-size calculation using the software G^*Power (statistics: one-way ANOVA for fixed effects). We based our calculations on the baseline and response cortisol values. Previous data showed that effects of the social environment on cortisol concentrations are large, e.g., effect sizes for large effects between f = 0.69–1.6 [e.g., (42, 54)]. To detect effects with f = 0.69 with a power of >80% a total sample size of at least 28 animals would be needed i.e., 7 per group. To detect large effects, already starting at f = 0.4–0.5 with a power of >80%, we decided to use 14 animals per group, resulting in a total sample size of 56 animals, i.e., 14 animals per group. The calculated sample size sometimes deviates considerably from the achieved sample size. In the case of sterilized animals, this is due to the surgery (see Section 2.1). Furthermore, at some time points too few males were born for four treatment groups. Thus, the sample size was the following: sterilized males: n = 4 (see Section 2.1 for reason for the small sample size), castrated males: n = 14; sham-operated males: n = 8; intact males: n = 10. Because the sample size was very small and measures violated assumptions for parametric tests and transformations failed to correct this, non-parametric statistics were used. Kruskal-Wallis tests were used to determine whether there were any treatment effects in testosterone, baseline and response values of cortisol between animals of the four groups. In case of significance, pairwise comparisons between castrated, sterilized, intact and sham-operated males were subsequently performed using Mann–Whitney U-tests (Holm-Bonferroni corrected). In general, two-tailed tests were used and significance was set at p < 0.05. Central tendency and variability for endocrine measures were expressed as medians and maximum and minimum values.

Testosterone concentrations did not differ between males before surgeries at the age of about 34 days. However, there was a significant difference of plasma testosterone concentrations between intact (n = 10), sham-operated (n = 8), castrated (n = 14) and sterilized (n = 4) males after surgery at each sample point (see Figure 1; for statistical details see Table 1). Post-hoc tests revealed that castrated males showed lower testosterone concentrations than intact, sham-operated and sterilized males (Figure 1; for statistical details see Table 1).

Cortisol concentrations were significantly different immediately before the males were transferred to the females in a new enclosure and 2 h later (intact males: n = 10; sham-operated males: n = 8, castrated males: n = 14, sterilized males: n = 4). Post-hoc tests showed that the values shortly before introduction into the groups were in castrated males higher than in sham-operated and castrated males (Figure 2; for statistical details see Table 2). Two hours after introduction, cortisol concentrations were significantly higher in castrated males than in intact, sham-operated and sterilized males (Figure 2; for statistical details see Table 2).

In the cortisol response test, baseline as well as response values of cortisol was determined. Before surgeries, no differences in cortisol baseline and response values were found (Figures 3, 4; for statistical details see Table 2). One week before the start of the experiment (i.e., surgeries were conducted, but males were not yet introduced to the females), differences in baseline values were found: post-hoc tests revealed higher values in castrated than in sham-operated males (Figure 3; for statistical details see Table 2). Baseline levels after surgery did not differ (Figure 3; for statistical details see Table 2). Regarding response values, 1 week before as well as 2 and 3 weeks after the start of the experiment significant differences could be determined. Post-hoc tests demonstrated that castrated males showed significantly higher C response values than intact, sham-operated and on two time points also than sterilized males (Figure 4; for statistical details see Table 2).

In females only cortisol baseline values were determined, i.e., blood samples were taken 1 day before as well as shortly, 1 and 3 weeks after housing them together with a castrated (n = 28), sham-operated (n = 16), sterilized (n = 8) or an intact male (n = 20). Only one minor difference could be found. Cortisol concentrations were significantly different before start of the experiment (Table 3). Post-hoc tests showed that females that were later housed with an intact male had significantly higher cortisol concentrations than females that were later housed with a sham-operated male (Mann-Whitney U-Test, two-tailed, U = 66.5, p < 0.05. No differences were observed at all time points after the females. Irrespective of whether they were housed with an intact, sham operated, sterilized or castrated or male.

As described in the introduction, baseline and response glucocorticoid values, are valid measures in stressful situations to assess animal welfare. In our study, cortisol baseline values did not differ between castrated males and intact, sham-operated and sterilized males (with one exception: before start of the experiment castrated males showed higher baseline values than sham-operated males). Thus, in non-challenging situations all males do not show any endocrinological signs of impaired welfare. Interestingly, however, castrated males showed higher cortisol response values after exposure to an acute stressor (novel environment, separation from their group), i.e., castrated males were more reactive in acute challenging situations than intact, sham-operated and sterilized males. This points to a less effective way to cope with stress situations in castrated than in non-castrated male guinea pigs. This could be a hint to a generalized impaired welfare in castrated males, at least in unpredictable and challenging situations. A similar finding could be shown in differently housed guinea pigs: Males housed singly showed a higher activation of the HPA system after an acute stressor than males housed together with a female. This indicates a less effective coping ability with stressful situations in guinea pigs kept alone compared to guinea pigs living in social groups (36).

Interestingly, no differences in cortisol concentrations were found in females living together with castrated, sterilized, sham-operated or intact males. This is contrary to our prediction and also to other species as for example mice. Garratt et al. (55) could show that female mice paired with castrated males had higher corticosterone concentrations (the main glucocorticoid in mice) than those females cohoused with sterilized males, thus indicating a higher stress response. These different results can possibly be explained by different forms of social organization, a different importance of males for females as well as a different importance of odors for communication.

Since males were castrated before sexual maturity in our study, testosterone levels were very low during adolescence. Interestingly, some studies could demonstrate that testosterone can be involved in organizing the HPA axis. In domestic guinea pigs, it was reported that the level of testosterone experienced during adolescence can shape cortisol responsiveness: if high testosterone concentrations are present during adolescence, low cortisol concentrations are released later in life during an acute stress situation. If low testosterone concentrations are present during adolescence, high cortisol concentrations are released later in life (56, 57). Thus, an inhibiting effect of testosterone on the cortisol response occurs during adolescence. It is very likely that this mechanism exists in a similar manner not only for domestic guinea pigs, but for other group-living animals (44, 58). Thus, in our study castrated males could show higher cortisol responsiveness during the cortisol response test in comparison to intact, sham-operated and sterilized males due to the mediating effects of testosterone on the HPA axis. The artificial alteration of the testosterone concentration through castration thus has far-reaching consequences for the stress reactivity of the animals. As this could be an indication of a lower coping capacity in acute stress situations, this could have implications for animal welfare.

Overall, our results point to a possibly impaired welfare in castrated animals. However, in our study we decided to focus on baseline and response levels of cortisol, which are well established indicators of animal welfare. For a broader, in depth assessment of animal welfare, other parameters must also be investigated in future studies. Besides cortisol, these could be other physiological parameters (e.g., heart rate or immunological measurements), but also ethological indicators (e.g., the extent of aggressive or stereotypic behavior) and the assessment of affective states [e.g., fear or pleasure; (59)]. Our study is limited due to the small sample size and the focus on cortisol baseline and response values. Nevertheless, the results strongly encourage further investigations on whether sterilization is not often an alternative to castration in terms of animal welfare when choosing a method to prevent uncontrolled reproduction.